Popular in Law

Law

Diversity and Social Mobility in the UK legal profession

Law Society I. Stephanie Boyce

Internet Law

The Law of AI for Good

University of San Diego Orly Lobel

Human Rights Law

Transforming Restorative Justice for Transitional Settings

University of Nottingham Kerry Clamp

Law

Surrey’s Silence: Subpart F and the Swiss Subsidiary Tax that Never Was

Boston University Steven Dean

Criminal Law

Prosecuting Misdemeanors

American University Jenny Roberts

Human Rights Law

Siege Starvation: A War Crime of Societal Torture

Tufts University Tom Dannenbaum

Corporate Law

Data Valuation and Law

University of Southern California Jordan Barry

Brave New World: Legislating for the Future of Driverless Cars

View more insights in Professional - Law - Criminal Law

View more figures from University of Leicester

View the page of Paul Almond

View the page of Sally Kyd

Read the Study

In January 2022, the Law Commission of England and Wales and the Scottish Law Commission on Automated Vehicles published their joint report on the regulation and safe deployment of automated vehicles. Paul Almond and Sally Kyd review the elements of the proposed regulations that employ the criminal sanction: the reimagining of the role of “criminal” responsibility and liability in relation to individual actors and “drivers”, and the development of a “regulatory” approach in relation to the organisational actors operating in this field.

Image courtesy of interviewee. August 8, 2024

Generating Workspace

Export as PDF

In this text, a discussion is held regarding the impact of automated vehicles on road safety and the legal implications surrounding them, specifically in the UK context. Sally and Paul, the speakers, explore how automated vehicles, if truly autonomous, could potentially reduce road harm by eliminating human error, which is a significant cause of road traffic collisions. They discuss the government’s tasking of the Law Commission to rethink legal frameworks for these vehicles.

Key points include:

1. **Conceptual Shift**: The transition from human-driven to automated vehicles necessitates re-evaluating legal liability—determining who is responsible when an automated vehicle is involved in an incident.

2. **User In Charge vs. Driver**: They introduce the concept of a "user in charge," who is different from a driver and is only responsible once the automated system requests control to be handed back. The implications of this distinction are explored concerning liability during transitional situations.

3. **Organizational Responsibility**: They emphasize that a significant portion of responsibility for road traffic offenses may shift to the operating company of the automated vehicles, which must also adhere to strict regulatory standards for safety and operations.

4. **Regulatory Framework**: The Law Commission proposed a need for new regulations that could accommodate automated vehicles, focusing on consumer protection, safety standards, and regulatory oversight.

5. **Challenges in Accountability**: They acknowledge concerns about bereaved families not achieving accountability or justice through the current legal framework when road incidents occur involving automated vehicles, as past precedents for human drivers may not apply.

6. **Future of Driving and Skills**: The speakers speculate on the potential loss of necessary driving skills among humans as reliance on automated systems increases, citing the need for ongoing education and adaptations in driver safety campaigning.

7. **Thresholds of Responsibility**: The discussion addresses the difficulties in defining and navigating the thresholds that determine when a user becomes a driver and how automated systems engage with unexpected situations on open roads.

Overall, the conversation highlights the need for thorough legal reform and regulatory frameworks to ensure road safety, maintain accountability, and effectively integrate automated vehicles into society. There is a call for further exploration of the implications of these technologies as they develop and become commonplace.

Language:

Export as PDF

Export as PDF
Here are five key points from the video regarding automated vehicles and the implications for law and liability:

1. **Reduction of Human Error**: The government believes that automated vehicles, or self-driving cars, have the potential to significantly reduce road traffic accidents caused primarily by human error. The effectiveness of fully automated vehicles in achieving this goal remains uncertain, as the technology is still under development.

2. **Regulatory Framework Development**: The Law Commission was tasked with exploring how existing laws (both criminal and civil) should be altered to accommodate automated vehicles and enhance the safety of their use on public roads. This project took over three years and was considered one of the most complex endeavors the Commission has undertaken.

3. **Liability Shift**: The introduction of automated vehicles necessitates a reassessment of liability for road traffic incidents. The law distinguishes between the "user in charge," who oversees the vehicle but does not drive it, and the automated self-driving entity (ASDE), which provides the necessary technology. This shift means that traditional liability associated with human drivers will need to adapt to the new ecosystem of automated driving.

4. **User in Charge vs. Driver Responsibilities**: When an automated vehicle is in operation, the user in charge is not expected to engage in driving tasks but must be available to take control if needed. If a control transition is demanded and the user fails to resume responsibility, they may become liable for driving offenses as they revert to the role of a driver.

5. **Calls for Accountability in Serious Incidents**: The video expresses concerns regarding accountability in the case of serious incidents involving automated vehicles, such as fatalities. Current legal frameworks for crimes like dangerous or careless driving may not apply directly to automated driving systems. This gap raises questions about how accountability can be ensured for automated vehicle manufacturers and operators. The Law Commission suggests corporate liability may be a solution in severe incidents, but overall accountability remains a critical challenge in the transition to automated vehicles.

Export as PDF
Here are some key quotes from the video:

1. **On the potential of automated vehicles:**
- "The government, at least, is of the view that automated vehicles can reduce road harm on the basis that if we can have truly automated vehicles, self-driving cars, that will remove the main cause of road traffic collisions, which is the human behind the wheel."

2. **On uncertainty regarding future developments:**
- "Whether that's actually possible going forward, I think is a question we don't yet know the answer to."

3. **On the aim of regulatory frameworks:**
- "The idea is that the Law Commission would put into place a regulatory framework that would allow that to happen."

4. **On redefining responsibility:**
- "Thinking about self-driving cars kind of brings into question a lot of the fundamental assumptions that tend to underpin thinking in particular about criminal responsibility for driving."

5. **On the roles of users in charge and operators:**
- "The term they use for that is a user in charge... they are supposed to be in a position to respond to an alert... The other actor who's really important... is what they call the ASDE or automated self-driving entity."

6. **On liability:**
- "...the driver is not driving the car and they don't bear the responsibilities we typically give to a driver as a user in charge."

7. **On the tasks of the law commission:**
- "Part of what's important... is that they decided that there ought to be an offense of allowing oneself to be carried in a vehicle without a user in charge where a user in charge is required."

8. **On future vehicles without a user in charge:**
- "Some vehicles they do envisage will be ones where no user in charge is required, which would be in some respects fully automated."

9. **On the challenges of accountability:**
- "...those who become bereaved by incidents on the roads won't necessarily get what they might be looking for in terms of criminal liability of either an individual or a company."

10. **On safety standards for automated vehicles:**
- "To be authorized a vehicle would have to be as safe, if not safer, than a competent and careful driver."

11. **On the significant changes in legal responsibility:**
- "The responsibility for, and the accountability for driving offenses... shifted much more to this regulatory learning regulatory pedagogy model than it is the traditional kind of road traffic policing model."

These quotes encapsulate the discussion surrounding the potential benefits and challenges of introducing automated vehicles and the regulatory changes necessary to accommodate them.

Export as PDF
Here are some articles that are similar in content or theme to the video , which discusses automated vehicles, legal implications of self-driving technology, and the responsibilities associated with their operation:

1. **"Regulating Driverless Cars: A Consideration of Policy and Legal Challenges"**
- This article explores the legal frameworks necessary for the safe integration of autonomous vehicles on public roads and discusses the potential liabilities for manufacturers and users.

2. **"The Future of Road Safety: Automated Vehicles and Legal Liability"**
- An analysis of how automated vehicles may change the landscape of road safety and the legal obligations of drivers and manufacturers in collision scenarios.

3. **"The Ethics of Self-Driving Cars: Navigating Liability and Accountability"**
- This paper examines ethical considerations arising from self-driving vehicles, including questions of legal responsibility and moral accountability in the event of accidents.

4. **"Towards an Autonomous Future: Challenges in Law for Self-Driving Vehicles"**
- A discussion on the implications of fully automated driving systems for current legal frameworks, focusing on necessary reforms and the role of regulatory bodies.

5. **"Can Autonomous Vehicles Eliminate Human Error? The Legal Framework for Automation"**
- This article debates the premise that autonomous vehicles can reduce accidents caused by human error, while outlining the legal responsibilities of both users and the companies producing these vehicles.

6. **"Implications of Self-Driving Cars on Liability Law"**
- A legal examination of how the existence of self-driving cars may shift liabilities from drivers to manufacturers and software developers in instances of accidents.

7. **"Smart Vehicles: The Intersection of Technology, Law, and Public Safety"**
- An overview of how increasing technological innovations in transportation require reevaluation of existing laws and regulations to ensure public safety and account for new risks.

8. **"Autonomous Vehicles: Law, Ethics, and Public Policy"**
- This paper explores public policy considerations and the ethical landscape surrounding the introduction of self-driving cars, including how law can adapt to these emerging technologies.

Each of these articles addresses critical discussions around the implications of automated vehicles, including legal, ethical, and safety issues that are raised in the video.

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.065 --> 00:02.565
The government at least, is of the view that

00:03.155 --> 00:06.685
automated vehicles can reduce road harm on the basis

00:06.835 --> 00:09.965
that if we can have truly automated vehicles,

00:09.965 --> 00:13.325
self-driving cars, that will remove the main cause

00:13.425 --> 00:17.325
of road traffic collisions on the roads, which is the human

00:17.425 --> 00:18.605
behind the wheel.

00:19.105 --> 00:22.405
Um, so human error, whether, whether

00:22.405 --> 00:25.185
that's actually possible going forward,

00:25.505 --> 00:28.785
I think is a question we, we don't yet know the answer to.

00:29.445 --> 00:33.305
Um, you know, every, every year we hear from the likes

00:33.305 --> 00:35.905
of Tesla that a fully automated vehicle is just round the

00:35.905 --> 00:38.345
corner, and, and that hasn't yet happened.

00:38.965 --> 00:42.885
Um, but that is, that is the idea that

00:42.915 --> 00:45.525
that automated vehicles will,

00:46.155 --> 00:48.485
have the potential at least to be safer than humans

00:48.595 --> 00:50.045
because,

00:50.435 --> 00:53.165
they will not be distracted in the way humans

00:53.235 --> 00:54.405
will, et cetera.

00:54.945 --> 00:58.485
And so that's why, the government tasked,

00:58.705 --> 01:02.925
the law commission with, a project,

01:03.385 --> 01:06.525
to think about how the law should be changed.

01:06.745 --> 01:08.645
Um, both the criminal law and civil law,

01:08.965 --> 01:11.365
although both of the authors of this piece, myself and

01:11.585 --> 01:13.125
and Paul are criminal lawyers,

01:13.585 --> 01:16.245
but how the law would need to be changed in order to,

01:16.585 --> 01:20.965
to accommodate automated vehicles on our roads, in order

01:20.995 --> 01:22.885
that I think, you know,

01:22.885 --> 01:26.045
they could support British industry in particular in,

01:26.145 --> 01:29.005
in bringing these, these vehicles onto the roads.

01:29.105 --> 01:32.355
Um, so that's the potential.

01:32.355 --> 01:35.755
Whether it can come to fruition, is another question.

01:35.775 --> 01:38.355
But the idea is that the Law Commission would put into place

01:38.635 --> 01:42.315
a regulatory framework that would allow that to happen.

01:42.735 --> 01:44.315
The Law Commission,

01:44.385 --> 01:46.635
took more than three years on this project,

01:46.645 --> 01:51.595
which I think is probably the, the most, well,

01:51.615 --> 01:53.235
the longest period

01:53.235 --> 01:55.075
that they've spent on, on that kind of project.

01:55.335 --> 01:57.235
And it was particularly interesting, as I say,

01:57.235 --> 02:00.035
because it covered, different areas of law.

02:00.035 --> 02:02.635
Usually the law commission is looking at one area of,

02:02.975 --> 02:04.795
of the law, whether it be criminal law,

02:04.795 --> 02:07.675
whether it be family law, and,

02:07.675 --> 02:10.355
and they have team specific to those areas of law.

02:10.415 --> 02:11.675
So this was a bit unusual

02:11.675 --> 02:13.795
because it was bringing together all sorts of,

02:14.095 --> 02:15.155
areas of law.

02:15.255 --> 02:17.955
Not only that, but it was a joint project

02:18.105 --> 02:19.635
with the Scottish Law Commission.

02:19.735 --> 02:21.835
So, you know, we are particularly interested in

02:21.835 --> 02:23.515
what the Law commission was saying,

02:23.695 --> 02:24.835
in England and Wales.

02:25.135 --> 02:26.915
Um, that's jurisdiction in which we work,

02:27.135 --> 02:29.115
but this is a joint report

02:29.185 --> 02:31.075
with the Scottish Law Commission as well.

02:31.535 --> 02:33.555
And so they published the outcome of

02:33.555 --> 02:35.435
that report in January, 2022.

02:36.015 --> 02:38.115
Um, and given the number of consultation papers,

02:38.165 --> 02:39.635
there had been, unsurprisingly,

02:39.635 --> 02:44.155
it was a very lengthy report, so over 300 pages now

02:44.295 --> 02:47.675
for anyone operating within the law, that's,

02:47.675 --> 02:50.875
that's a very large document to have to get your head round.

02:51.295 --> 02:55.155
Um, and so I thought it would be useful to, to, to try

02:55.155 --> 02:58.275
and, bring out for criminal lawyers in particular,

02:58.535 --> 03:01.685
the most important aspects of, of that report.

03:01.745 --> 03:03.845
And I asked Paul to join me because

03:04.085 --> 03:07.645
although I've got expertise in road traffic law and,

03:07.665 --> 03:09.605
and other criminal offenses,

03:09.785 --> 03:12.965
he has much more expertise in, in,

03:13.685 --> 03:15.485
regulatory frameworks and,

03:15.585 --> 03:18.405
and the liability of companies in particular,

03:18.405 --> 03:20.445
which is particularly relevant to, to

03:20.445 --> 03:21.605
what the law commission had said.

03:21.835 --> 03:26.525
Part of what's important about it as a, a topic area is

03:26.525 --> 03:30.645
that thinking about self-driving cars kind of

03:31.305 --> 03:35.005
brings into question a lot of the fundamental assumptions

03:35.005 --> 03:38.485
that tend to underpin thinking in particular about criminal

03:38.785 --> 03:40.365
responsibility for driving.

03:41.425 --> 03:45.925
You end up having to kind of reset the, the, the,

03:46.625 --> 03:49.005
the rules and the frames of reference for how

03:49.305 --> 03:51.565
and when you think about questions of liability.

03:52.305 --> 03:56.645
So the report itself is talking about

03:57.035 --> 03:59.485
automated vehicles,

03:59.705 --> 04:02.965
and in particular an automated vehicle is one that,

04:03.465 --> 04:05.205
and the term that they, the,

04:05.755 --> 04:07.405
they're talking about is it is capable

04:07.545 --> 04:10.565
of performing the dynamic driving task by itself.

04:11.465 --> 04:15.325
So this is different than any kind of, you know,

04:15.515 --> 04:18.845
lane assist or driver assist software

04:18.845 --> 04:21.205
that might already exist or be in, in use.

04:22.025 --> 04:25.245
Uh, an automated vehicle is a, a vehicle, a car

04:25.245 --> 04:27.965
that drives itself independent of the driver.

04:29.265 --> 04:30.685
And,

04:33.275 --> 04:37.255
so in thinking about how they define the,

04:37.985 --> 04:41.335
who's responsible for what in this field, the two terms

04:41.365 --> 04:44.735
that they talked about, the two actors that they started to

04:45.265 --> 04:49.095
think about responsibility for were, if

04:50.155 --> 04:53.355
we, if they're not driven by a driver, then

04:53.535 --> 04:56.915
who is the person in the car who's in charge of the car

04:56.915 --> 04:58.155
or responsible for the car?

04:58.455 --> 05:00.875
The term they use for that is a user in charge.

05:01.455 --> 05:03.435
So a user in charge is not a driver,

05:04.065 --> 05:05.955
they're not doing the driving task,

05:06.735 --> 05:10.955
and they are not necessarily meant to be paying attention

05:11.655 --> 05:14.195
to the driving of the vehicle while it's driving,

05:14.935 --> 05:18.035
but they are supposed to be in a position to respond

05:18.695 --> 05:20.915
to an alert or a request

05:21.095 --> 05:23.195
or a need to take control of the vehicle.

05:23.735 --> 05:25.915
And obviously as the person in charge of the vehicle,

05:25.975 --> 05:28.235
the user in charge, they're also responsible for

05:29.265 --> 05:30.955
framework issues around the vehicle,

05:31.025 --> 05:33.715
like its road worthiness and, and so on and so forth.

05:34.695 --> 05:38.755
And the other actor who's really important in the field are

05:38.755 --> 05:42.835
what they call the a SD or automated self-driving entity.

05:43.595 --> 05:45.415
So this is the operating company

05:46.125 --> 05:48.015
that will be providing the software

05:48.715 --> 05:51.695
and hardware that allows the vehicle to drive itself and,

05:52.075 --> 05:55.215
and running it through their software models,

05:55.215 --> 05:57.965
through their support systems and infrastructure.

05:58.985 --> 06:03.005
So traditionally, when we think about responsibility for

06:04.035 --> 06:06.725
road traffic and accidents on the road,

06:06.825 --> 06:11.405
or, you know, deaths, collisions, anything else like that

06:12.595 --> 06:14.315
that sits with the driver,

06:14.495 --> 06:17.355
unless you are talking about something much further

06:17.425 --> 06:20.515
that back down the food chain of, of a manufacturing fault

06:20.575 --> 06:23.475
or something similar in this situation,

06:24.495 --> 06:26.475
the driver is not driving the car

06:26.495 --> 06:29.475
and they don't bear the responsibilities we typically give

06:29.475 --> 06:31.755
to a driver as a user in charge.

06:31.885 --> 06:34.715
Their responsibilities are much more narrowly defined

06:34.715 --> 06:38.555
and framed and much more responsibility

06:39.015 --> 06:42.955
for the operation sits with the A SDE, the operator,

06:43.535 --> 06:46.715
the the organizational actor that provides the vehicle.

06:47.335 --> 06:51.075
So they're having to redevelop new ways of thinking about

06:51.785 --> 06:56.355
when those two new kinds of actor become responsible

06:56.705 --> 06:57.835
that weren't there before.

06:57.935 --> 06:59.675
And that's what was really tricky

06:59.775 --> 07:02.235
and challenging for the law commission to do

07:02.415 --> 07:07.075
The user in charge, is somebody, who

07:07.735 --> 07:11.435
is not expected when the,

07:11.545 --> 07:14.675
automated driving system is in control of the, the, the car.

07:14.695 --> 07:17.435
So when the, the vehicle itself is, is,

07:17.465 --> 07:21.035
engaged in the dynamic driving task, the user in charge

07:21.685 --> 07:23.875
isn't expected to do anything in particular.

07:23.875 --> 07:25.835
They can pretty much do what they like. Okay.

07:26.375 --> 07:30.995
But, as soon as there is that transition demand,

07:31.665 --> 07:34.395
once that the, the period comes to an end,

07:34.415 --> 07:36.195
so the suggestion is that might be 10 seconds

07:36.265 --> 07:38.235
between when the vehicle says you need

07:38.235 --> 07:39.555
to take back control of the vehicle.

07:40.095 --> 07:43.395
Um, once that happens, that user in charge then kind

07:43.395 --> 07:45.515
of metamorphosis in back into a driver.

07:46.825 --> 07:49.645
So once they are a driver, that same person

07:49.745 --> 07:52.885
who was the user in charge is then liable

07:53.105 --> 07:57.325
for any driving offenses that they might commit.

07:57.985 --> 07:59.165
And one of the things that the,

07:59.165 --> 08:01.685
the law commission were having to think about was, well,

08:01.685 --> 08:03.685
what happens if, if somebody who is in the vehicle

08:04.035 --> 08:07.045
with a responsibility of the user in charge fails

08:07.045 --> 08:10.845
to take back responsibility should recreate a new offense

08:11.115 --> 08:14.605
that would, make them liable for failing to do that?

08:14.665 --> 08:16.885
And they decided that, that they weren't going to do that,

08:17.345 --> 08:19.845
that actually because they then became a driver,

08:19.915 --> 08:21.205
they would just be liable

08:21.205 --> 08:23.365
for whatever driving offenses they then,

08:23.395 --> 08:25.045
committed at that period in time.

08:25.225 --> 08:28.765
So if they haven't taken back, control, then presumably,

08:29.585 --> 08:31.845
they will be falling below the standard

08:31.845 --> 08:33.405
of a competent and careful driver.

08:33.825 --> 08:36.365
And so would be committing automatically, if you like,

08:36.365 --> 08:38.725
the offensive care that's driving a potentially dangerous

08:38.725 --> 08:40.285
driving depending on the circumstances.

08:40.785 --> 08:42.565
So that was one of the things they looked at.

08:42.875 --> 08:47.005
Another offense in terms of individuals is, is what if,

08:47.265 --> 08:49.405
you know, you've got, a group

08:49.405 --> 08:53.725
of people there is this vehicle that is capable of, of,

08:54.455 --> 08:57.285
being driven by itself,

08:57.945 --> 09:00.125
but it does require a user in charge.

09:00.825 --> 09:03.445
Um, and none of them want to take that responsibility.

09:03.625 --> 09:05.605
So, so a bunch of people get in the car

09:05.605 --> 09:08.205
and nobody actually, takes responsibility

09:08.265 --> 09:09.685
for sitting in the driving seat.

09:10.265 --> 09:14.605
Um, then they have decided that there ought to be,

09:14.705 --> 09:19.445
an offense of allowing oneself to be,

09:19.635 --> 09:22.405
carried in a vehicle without a user in charge

09:22.535 --> 09:24.805
where a user in charge is required.

09:24.905 --> 09:26.965
And I suppose what we should mention is

09:26.965 --> 09:29.845
that the law commission here had to envisage two types

09:29.865 --> 09:32.405
of vehicle, to be able to adapt to,

09:32.585 --> 09:34.845
to the way in which technology will be developed.

09:34.905 --> 09:36.965
So, so in the first instance, they're,

09:36.965 --> 09:39.085
they're thinking about vehicles that probably

09:39.655 --> 09:43.725
won't do the full, full jour journey itself in terms of

09:44.345 --> 09:46.405
the dynamic driving task that, that, you know, a,

09:46.485 --> 09:48.645
a driver may well have to begin the journey,

09:49.275 --> 09:52.885
then they can hand over control to the automated,

09:53.805 --> 09:54.815
driving system.

09:55.235 --> 09:57.895
And then at some point that system will want hand back,

09:58.435 --> 09:59.615
responsibility.

09:59.635 --> 10:01.455
And, and that's where you have a user in charge.

10:02.115 --> 10:06.015
Um, but some vehicles they do envisage in future will,

10:06.075 --> 10:09.895
be vehicles where no user in charge is required,

10:09.945 --> 10:12.815
which would be in some respects fully automated.

10:13.275 --> 10:16.255
But in that, in, in those cases, there would need to be

10:16.975 --> 10:19.935
somebody, employed by the A SDE.

10:20.275 --> 10:22.695
Um, so the company that runs the vehicle who,

10:22.755 --> 10:25.895
who would then, have oversight of the vehicle.

10:26.195 --> 10:29.655
So it's, that's not the same as remote driving, but,

10:29.715 --> 10:31.615
but they would have some control

10:31.615 --> 10:33.135
of the vehicle from a distance.

10:33.155 --> 10:37.415
And those, those vehicles, would be ones, where

10:37.645 --> 10:39.175
that offense of, of their failing

10:39.235 --> 10:42.295
to be a user in charge would, would not be committed

10:42.295 --> 10:43.895
because if you don't need a user in charge,

10:43.895 --> 10:45.655
then people could just get in the car and,

10:45.955 --> 10:47.935
and, carry on.

10:47.995 --> 10:52.055
So, that, that's one of the things, I mean, it, it,

10:52.205 --> 10:53.415
this is part of the issue here.

10:53.415 --> 10:56.175
It's so complex that the, the having to envisage

10:56.205 --> 10:57.375
what might be needed

10:57.875 --> 10:59.855
and also then the confusion that

10:59.855 --> 11:01.575
that can create for human beings.

11:01.715 --> 11:03.575
So, so this question of, you know, how,

11:03.715 --> 11:05.655
how would somebody know when they got in a vehicle,

11:05.925 --> 11:08.695
whether it was one that required a user in charge

11:08.955 --> 11:11.695
or not, do they know that,

11:11.695 --> 11:13.575
they might be committing an offense if,

11:13.595 --> 11:14.935
if there isn't a user in charge

11:14.955 --> 11:18.095
and they just let the, the vehicle drive itself?

11:18.675 --> 11:19.775
Um, but the benefit

11:19.835 --> 11:23.135
of a no user in charge vehicle might be in future

11:23.205 --> 11:24.775
that you could have people who would

11:24.775 --> 11:28.135
otherwise be incapable of driving perhaps, for,

11:28.155 --> 11:29.375
for reasons of disability.

11:29.995 --> 11:31.215
Um, and, and,

11:31.315 --> 11:33.455
and I think that's one thing as well as the,

11:33.455 --> 11:34.655
the road safety aspect.

11:34.995 --> 11:38.095
Um, there's this idea that we do want to be able to provide

11:38.315 --> 11:42.495
for other members of society who can't at the moment,

11:42.685 --> 11:45.895
rely on, on motor vehicles for, for,

11:46.075 --> 11:48.855
for transport in the same way as they, they might be able

11:48.855 --> 11:50.975
to if there truly was automated vehicles.

11:51.275 --> 11:52.735
One of the things that

11:52.735 --> 11:54.925
The proposals do is shift a lot

11:54.925 --> 11:57.765
of onus away from driver responsibility

11:58.195 --> 12:00.805
because you don't have a driver in the way

12:00.805 --> 12:03.405
that we traditionally or currently conceive of it.

12:03.865 --> 12:06.765
So the question then becomes, where does a lot of

12:06.765 --> 12:08.285
that liability go to?

12:08.785 --> 12:12.165
And the answer tends to be the operating organizations,

12:13.185 --> 12:16.565
the A SDE, the self-driving entity organization

12:16.565 --> 12:20.365
that is running or utilizing, providing the technology.

12:20.905 --> 12:23.125
And there are really three kinds of offenses

12:23.125 --> 12:27.245
that the law commission had envisaged relating to these,

12:27.425 --> 12:28.805
automated vehicles.

12:29.425 --> 12:32.765
The first ones are kind of around the sale and manufacture

12:32.905 --> 12:36.205
and, and advertising of these vehicles that they should

12:36.995 --> 12:39.405
only claim to do what they actually do.

12:39.435 --> 12:40.925
They shouldn't advertise themselves

12:40.945 --> 12:43.245
as self-driving if they're not self-driving and so on.

12:43.245 --> 12:45.325
So these are kind of a bit like consumer protection

12:45.355 --> 12:46.365
type of offenses.

12:47.225 --> 12:50.325
It probably would be fair to say that I think the,

12:50.435 --> 12:53.725
automated vehicle industry has not necessarily always

12:54.605 --> 12:59.405
observed these principles, up to now in terms of,

12:59.785 --> 13:01.405
the way that it sells or,

13:01.425 --> 13:03.005
or advertises,

13:04.095 --> 13:06.445
quasi autonomous vehicles up to this point.

13:06.665 --> 13:09.645
So that kind of motivated the law commission

13:09.645 --> 13:11.205
to feel like they needed offenses

13:11.225 --> 13:14.045
to draw a bright line distinction between

13:14.775 --> 13:16.245
truly self-driving cars

13:16.705 --> 13:19.005
and cars that are not truly self-driving,

13:19.345 --> 13:22.685
but that will take over crew cruise control,

13:23.675 --> 13:27.085
lane assistance, you know, all of these other kind

13:27.085 --> 13:29.005
of adjustments that that can help the driver.

13:29.865 --> 13:33.485
The second kind of offenses are ones about, I guess kind

13:33.485 --> 13:36.685
of regulatory, openness

13:37.145 --> 13:38.765
and cooperation.

13:41.455 --> 13:44.675
The law commissioner proposed introducing two new regulatory

13:44.835 --> 13:48.915
bodies, one of which would be about the manufacture,

13:49.295 --> 13:53.075
the safety standards, the provision of automated vehicles.

13:53.415 --> 13:56.795
The second one would be a kind of in use regulator to deal

13:56.795 --> 13:59.675
with the way that they operate day-to-day on the street,

13:59.675 --> 14:02.955
and if there are problems associated with them tending to,

14:04.015 --> 14:08.275
you know, not respond to particular hazards or issues

14:08.535 --> 14:10.635
or known problems that come up

14:10.775 --> 14:13.075
or investigating accidents when they occur.

14:13.735 --> 14:16.915
And the law commission introduced new offenses relating to

14:17.635 --> 14:20.475
a kind of duty of candor on the companies to

14:21.085 --> 14:23.595
share information with those regulators,

14:24.215 --> 14:27.955
to provide them details, to allow them to see,

14:28.085 --> 14:30.115
everything they need to know about the code

14:30.115 --> 14:31.355
and the operating system,

14:31.735 --> 14:34.635
and to share details crucially of accidents,

14:34.735 --> 14:36.195
or collisions when they occur.

14:37.135 --> 14:41.075
Um, again, probably something that, it, it's fair

14:41.075 --> 14:43.795
to say, the,

14:44.755 --> 14:47.075
emerging automated vehicle industry hasn't always

14:47.075 --> 14:51.445
necessarily, been, it was super keen to do in practice,

14:51.705 --> 14:54.365
in every eventuality,

14:55.625 --> 14:57.485
the most important areas are those related

14:57.505 --> 14:59.045
to road traffic infractions,

14:59.045 --> 15:00.845
because this is where they're plugging the gap

15:01.095 --> 15:04.005
where a driver would normally have been responsible,

15:04.785 --> 15:06.005
and they kind of

15:06.955 --> 15:09.805
have a model which is based upon the principle of

15:10.765 --> 15:12.005
learning from adverse events.

15:12.625 --> 15:16.845
So the idea should be that the, this duty

15:16.945 --> 15:21.045
to share information is there, self-driving companies have

15:21.045 --> 15:23.685
to provide information about a, a collision

15:23.745 --> 15:25.005
or an incident or an event.

15:25.705 --> 15:29.005
And then the regulator's primary

15:29.925 --> 15:32.525
interest would be in using that information

15:32.665 --> 15:35.405
to iron out the problems that have led to that issue.

15:35.705 --> 15:37.765
Is it about recognizing bus lanes?

15:37.765 --> 15:41.325
Is it about being able to identify the difference between,

15:41.945 --> 15:45.405
a, you know, an object

15:45.545 --> 15:49.085
and a push chair that things that in practice, you know,

15:49.695 --> 15:52.525
could potentially be serious problems you want

15:52.525 --> 15:55.445
to learn when things go wrong, make adjustments

15:55.465 --> 15:56.805
and make collect collections.

15:57.265 --> 15:59.925
So it moves away very much from the kind of model

15:59.925 --> 16:01.765
that we'd tend to be used, used to

16:02.305 --> 16:06.165
of getting a speeding ticket or a, a a a fine

16:06.345 --> 16:09.765
or something for, exceeding a speed limit

16:09.825 --> 16:13.565
or being in a bus lane or, or whatever else it might be.

16:14.505 --> 16:18.085
And taking that approach in relation to those kind of,

16:18.085 --> 16:21.685
what you might call factual infractions is probably not

16:22.595 --> 16:23.685
that controversial.

16:23.945 --> 16:26.205
An issue where it's more controversial

16:26.725 --> 16:30.085
probably is in relation to incidents that lead

16:30.085 --> 16:31.925
to more serious consequences.

16:32.905 --> 16:37.005
So where, a a, a death has been caused,

16:37.895 --> 16:40.355
the existing offense is causing death by dangerous driving,

16:40.355 --> 16:41.915
causing death by carless driving.

16:42.185 --> 16:46.915
They wouldn't map across to these organizational entities

16:47.185 --> 16:49.795
because they're not doing the driving,

16:50.335 --> 16:51.875
the car is driving itself,

16:52.415 --> 16:55.035
so they haven't caused the death by dangerous driving.

16:55.855 --> 16:59.945
And by making, that kind of

17:00.625 --> 17:02.825
emphasis on learning from those adverse events,

17:04.085 --> 17:07.305
the likelihood might be that you end up with a sort of gap

17:07.395 --> 17:12.025
where some, a car doesn't do something it's supposed to do,

17:12.735 --> 17:14.705
there's a problem that should have been picked up

17:14.705 --> 17:17.435
that perhaps that wasn't somebody been

17:17.655 --> 17:18.875
plowed into and killed.

17:19.815 --> 17:23.475
And the primary approach of the, the model is liable

17:23.495 --> 17:26.555
to be about, well, what can we learn from this

17:26.775 --> 17:28.595
and how do we put it right for the future?

17:29.885 --> 17:31.635
There tend to be a lot of public interest

17:31.655 --> 17:34.315
and obviously interest from victims, families and

17:34.315 --> 17:38.475
and relatives in, in a, in securing some kind of redress,

17:38.905 --> 17:42.355
some kind of accountability, in those situations

17:42.355 --> 17:43.755
because,

17:43.975 --> 17:46.435
we would be talking about probably higher levels of,

17:46.935 --> 17:49.035
of culpability associated with

17:49.125 --> 17:51.045
what had happened rather than just, well,

17:51.505 --> 17:53.165
it was supposed not to go into that lane

17:53.165 --> 17:54.285
and it did go into that lane

17:54.285 --> 17:57.325
and now somebody's dead, dear, we'll learn from it.

17:57.505 --> 18:00.325
So there's a gap there in relation

18:00.325 --> 18:03.165
to potential liability when things go very seriously

18:03.165 --> 18:04.285
wrong in that kind of way.

18:04.985 --> 18:06.925
But there might be some areas where

18:07.685 --> 18:10.045
existing legal measures do plug some of that gap.

18:10.665 --> 18:13.085
Um, one example of that, for instance, would be

18:13.655 --> 18:16.085
there is a criminal offense of corporate manslaughter.

18:16.505 --> 18:20.045
It applies to any corporate organization that,

18:20.385 --> 18:22.525
causes a death that which results from

18:23.045 --> 18:25.725
a management failure in the way that it organizes

18:25.745 --> 18:26.845
or runs its business.

18:27.225 --> 18:29.725
And this will apply to any organization,

18:29.725 --> 18:31.045
whether it's a building firm

18:31.505 --> 18:34.925
or whether it's an automated self-driving entity,

18:35.155 --> 18:36.885
operating self-driving cars.

18:37.705 --> 18:41.005
So if you could show that management failures in the way

18:41.005 --> 18:43.645
that these organizations ran their business

18:43.865 --> 18:46.485
and managed the cars that they

18:46.765 --> 18:50.005
provided, the vehicles that were out there on the road,

18:51.145 --> 18:53.405
it seems that there probably would potentially be

18:53.995 --> 18:57.165
some avenue for corporate criminal liability.

18:57.665 --> 19:01.805
Now, in practice, in any regulatory setting, those kind

19:01.805 --> 19:05.405
of offenses tend to be used very rarely as a last resort

19:05.425 --> 19:07.685
and only in the most egregious kind of cases.

19:08.065 --> 19:11.845
So it's unlikely that we would see kind of vast numbers

19:11.945 --> 19:15.325
of these, of these cases coming before court.

19:15.545 --> 19:19.725
But primarily the, the responsibility for,

19:20.545 --> 19:24.565
and the accountability for driving offenses as we tend

19:24.565 --> 19:27.045
to think about them, it shifted much more

19:27.045 --> 19:28.485
to this regulatory learning

19:29.125 --> 19:32.405
regulatory pedagogy model than it is the traditional kind

19:32.405 --> 19:35.445
of road traffic policing model that we've seen up till now.

19:35.585 --> 19:36.685
And that's a trade off

19:36.685 --> 19:39.405
that I think the law commission said you kind of have

19:39.405 --> 19:43.565
to make, if you want to introduce, a, a reliable

19:45.295 --> 19:49.185
high, a reliable system in a, in a high risk novel

19:49.695 --> 19:51.105
technological setting.

19:51.415 --> 19:53.345
It's the same sort of approach, for instance, that

19:53.535 --> 19:56.465
that has traditionally been used in civil aviation,

19:57.145 --> 19:59.465
industry learn from the mistakes.

20:00.125 --> 20:02.905
But as I said, there are trade offs in accepting

20:02.905 --> 20:04.025
that that's the way you wanna go.

20:04.205 --> 20:06.625
We have a number of conclusions, and,

20:06.645 --> 20:08.985
and I think even since we published the paper,

20:09.165 --> 20:11.265
things have obviously developed.

20:11.885 --> 20:15.665
Um, and I think, you know, a lot of

20:15.665 --> 20:18.465
what we've said still remains true.

20:19.125 --> 20:23.825
Um, we, we've just said that, what one of the trade-offs

20:24.165 --> 20:29.025
for, for, for what was proposed, is that, those

20:29.165 --> 20:30.745
who become bereaved by

20:31.345 --> 20:35.065
instance on the roads won't necessarily, get

20:35.065 --> 20:37.065
what they might be looking for in terms of

20:37.745 --> 20:40.865
criminal liability of either an individual or,

20:40.925 --> 20:41.985
or, or a company.

20:42.125 --> 20:46.425
And I think, that is, that is gonna be a big challenge

20:46.535 --> 20:49.085
that, that that government and, and,

20:49.085 --> 20:52.005
and policy makers will have to face is, is responding

20:52.545 --> 20:54.525
to the needs of bereaved families.

20:54.725 --> 20:57.485
I mean, it's already the case that under the present system,

20:57.985 --> 21:01.845
it's felt by many that, that the, the, the needs

21:01.845 --> 21:04.685
of bereaved aren't met by the criminal justice system.

21:04.745 --> 21:07.605
And, and, and I think managing expectations there is,

21:07.625 --> 21:10.325
is going to be really important.

21:10.865 --> 21:14.685
Um, but it's this, kind of conceptual leap from,

21:14.685 --> 21:19.085
from thinking about things a, a as we have done a as humans

21:19.345 --> 21:23.645
to, you know, judging humans behind the wheel

21:23.765 --> 21:27.845
of a vehicle that is potentially very dangerous,

21:28.505 --> 21:32.965
to handing over responsibility to these, asds,

21:33.465 --> 21:37.285
and, and having to trust in to some extent

21:37.285 --> 21:40.965
that the technology can do a better job than human beings.

21:41.465 --> 21:44.565
Um, but I think, I, one of the,

21:44.625 --> 21:47.485
the disappointments I have, coming outta that

21:47.485 --> 21:50.005
and one of the recent developments is we, we, one

21:50.005 --> 21:52.325
of our conclusions is, the point

21:52.355 --> 21:55.645
that the law commission was unable to make the decision as

21:55.645 --> 21:57.565
to how safe a vehicle would have to be

21:57.565 --> 22:00.605
before it was given authorization to, to,

22:00.665 --> 22:05.485
to be on the roads, as,

22:05.905 --> 22:07.445
an automated driving system.

22:08.695 --> 22:10.915
And, they kind of said, well, that's

22:10.915 --> 22:12.115
for government to decide.

22:12.775 --> 22:15.715
Um, and they looked at a number of options.

22:16.185 --> 22:18.875
Well, of course we now are in a position where government,

22:19.575 --> 22:20.915
has put

22:20.915 --> 22:24.755
before parliament an act, which actually puts into practice,

22:25.375 --> 22:28.235
puts on on the statute book, most

22:28.295 --> 22:30.715
of the recommendations of the law commission,

22:30.855 --> 22:32.355
and goes further in the sense

22:32.355 --> 22:33.995
that it does answer that question as well.

22:34.055 --> 22:39.035
So under section two of the automated vehicles at, 2024,

22:39.495 --> 22:42.795
which is about the statements of safety principles,

22:42.895 --> 22:44.795
it makes the point that,

22:45.855 --> 22:49.555
to be authorized a vehicle would have to be as safe,

22:49.775 --> 22:52.475
if not safer, than a competent and careful driver.

22:53.335 --> 22:55.755
Um, and I think there's a problem there as,

22:55.775 --> 22:59.635
as we comment in the article, that that, that kind

22:59.635 --> 23:03.555
of perpetuates this idea that we can humanize, these,

23:03.725 --> 23:08.065
these robots if you like, which it isn't,

23:08.215 --> 23:11.585
helping I think with, with, with that conceptual leap that

23:11.585 --> 23:15.505
that, that we as society need to make, that the point is

23:15.625 --> 23:18.185
that these, these don't learn like human beings.

23:18.185 --> 23:20.345
They're not gonna be distracted like human beings, but,

23:20.365 --> 23:22.585
but they, they learn as a fleet.

23:22.655 --> 23:26.785
Machine learning is about each, element of a, a fleet

23:26.785 --> 23:29.385
of vehicles being able to learn from all the others

23:29.385 --> 23:30.385
that are on the roads.

23:30.765 --> 23:35.065
Um, and, and I, I personally don't feel that that that test

23:35.085 --> 23:37.585
of it being as safe as a competent

23:37.585 --> 23:40.905
and careful driver works in any sense of the word.

23:41.025 --> 23:44.865
I don't know how, the, the decision maker,

23:45.235 --> 23:49.005
will ever decide whether a vehicle is as safe as a competent

23:49.005 --> 23:52.645
and careful driver, which leaves open, you know, the,

23:52.645 --> 23:56.125
the other question is, is we now have this statute,

23:56.175 --> 24:00.165
which puts into practice many of the recommendations,

24:00.945 --> 24:03.925
but it's still reliant on a lot of secondary legislation

24:03.955 --> 24:05.365
that is yet to be brought in

24:05.365 --> 24:08.045
that would create the necessary regulatory bodies

24:08.555 --> 24:10.285
that would make these decisions.

24:10.665 --> 24:12.885
And so there's still a lot of work to be done there,

24:13.085 --> 24:15.445
I think, in, in, in that regard.

24:15.985 --> 24:19.045
Um, so I think those, those are some

24:19.045 --> 24:20.925
of the main conclusions.

24:20.925 --> 24:24.725
And I think the other one is that, you know,

24:24.775 --> 24:26.085
these things will take time.

24:26.265 --> 24:29.525
And, and my concern from the perspective of somebody who has

24:30.165 --> 24:33.045
researched fatalities on the roads and, and how we can try

24:33.045 --> 24:36.925
and reduce road date danger in the, long run is that,

24:36.995 --> 24:39.805
that there is going to be this transition period, not,

24:39.805 --> 24:44.725
not the 10 seconds of handing over,

24:44.875 --> 24:47.005
control of a vehicle back to a human being

24:47.785 --> 24:48.885
in a particular instance,

24:48.885 --> 24:51.725
but a transition period over years, if not decades,

24:52.375 --> 24:55.685
where you will have on the roads, both ordinary,

24:55.715 --> 24:59.205
what we consider ordinary vehicles where you have a driver,

24:59.605 --> 25:03.285
a human behind the wheel who might be making mistakes,

25:03.425 --> 25:06.365
but at the same time, you will also have these automated

25:06.365 --> 25:10.725
vehicles on the roads, engaging in machine learning,

25:11.305 --> 25:12.685
and which may

25:12.685 --> 25:14.445
or may not be handing back control to,

25:14.785 --> 25:16.285
to a user in charge in the vehicle.

25:17.065 --> 25:21.645
And I think, you know, from my perspective in terms of,

25:22.065 --> 25:25.205
the campaigns that I see about trying to,

25:25.345 --> 25:26.725
to improve road safety

25:26.825 --> 25:29.925
or reduce danger on the roads, it's mainly about

25:30.605 --> 25:33.005
focusing on the driver and making sure that they are fit

25:33.465 --> 25:37.925
to be a driver and am I, you know, the, the worry that,

25:38.385 --> 25:41.365
the skills of drivers are deteriorating, the suggestion

25:41.365 --> 25:42.765
that we should have graduated driving

25:42.885 --> 25:44.405
licenses, et cetera, et cetera.

25:44.405 --> 25:47.285
And that kind of all falls by the wayside if we're then,

25:47.765 --> 25:50.365
focusing instead on handing everything over to an

25:51.445 --> 25:56.125
a computer, if you like, will human beings retain the skill

25:56.125 --> 25:59.885
that they have, that is needed to drive vehicles

26:00.065 --> 26:02.285
and to be, a user in charge

26:02.545 --> 26:05.005
who might be given the responsibility of, of,

26:05.005 --> 26:06.405
of driving the fu

26:06.405 --> 26:09.085
driving the vehicle at any particular time? I think

26:09.545 --> 26:10.885
One of the interesting things

26:10.905 --> 26:15.125
and the most challenging things within the, the,

26:15.465 --> 26:16.725
the reform proposals

26:16.725 --> 26:18.965
and then the legislation that has followed

26:19.625 --> 26:22.205
and the implications it has are all about,

26:22.365 --> 26:24.125
I guess the concept of thresholds

26:24.125 --> 26:25.805
and when thresholds get crossed.

26:25.945 --> 26:29.165
So there's a threshold between being a user in charge

26:29.185 --> 26:31.405
of a vehicle and then becoming a driver,

26:31.985 --> 26:34.445
and how, you know, how much time you are given,

26:35.025 --> 26:38.525
how fair it is in those circumstances to expect somebody

26:38.545 --> 26:40.965
to suddenly become the driver in charge

26:41.105 --> 26:44.115
and drive to the standard of a, a,

26:44.315 --> 26:45.835
a competent and careful driver.

26:46.375 --> 26:49.275
If the car's already going off a cliff at the point

26:49.275 --> 26:50.635
that control is handed back,

26:51.105 --> 26:52.875
there's not much you can do as a driver.

26:53.695 --> 26:55.395
But there are also these thresholds

26:55.545 --> 26:56.955
that we encounter all the time,

26:57.175 --> 27:00.195
and automated vehicles have to, we'll have

27:00.195 --> 27:02.355
to deal with those.

27:02.495 --> 27:06.035
So, you know, as an example on my commute into work, I kind

27:06.035 --> 27:07.315
of come past the rugby ground

27:07.315 --> 27:09.875
and if there's a big crowd going in,

27:10.245 --> 27:13.355
there might be pedestrians in the road waving flags on their

27:13.375 --> 27:15.435
way, the car has to adjust to those.

27:15.495 --> 27:18.275
But if it can't, if it sees something it doesn't expect

27:18.415 --> 27:21.555
or doesn't know, it will either have to respond

27:21.575 --> 27:22.875
by giving control to me

27:23.375 --> 27:25.795
or it will do what automated vehicles,

27:26.195 --> 27:29.475
I think primarily tend to do when something goes wrong

27:29.535 --> 27:31.955
or when they try to hand back control and they can't.

27:31.955 --> 27:33.915
Or there's a system failure, which is

27:33.915 --> 27:35.035
that they will stop themselves.

27:35.465 --> 27:37.955
They stop when it's not safe to continue.

27:38.855 --> 27:40.755
So they have to deal

27:40.755 --> 27:42.515
with stuff they're not expecting to see.

27:42.575 --> 27:45.115
And I think the phrase we use in the,

27:45.335 --> 27:48.475
or in the, in the paper is about driving

27:48.655 --> 27:51.475
and the roads being an open system, not a closed one.

27:51.505 --> 27:54.835
It's not self-contained, like, you know,

27:54.835 --> 27:55.835
maybe the railways

27:55.835 --> 27:58.715
or a monorail would be, everything's on the tracks

27:58.715 --> 28:01.515
and it's all contained in that system, and you can automate

28:01.515 --> 28:03.435
and control it in the real world.

28:03.775 --> 28:05.235
You have stuff coming in left, right,

28:05.235 --> 28:08.995
and center permeating it, that that has to be accounted for.

28:09.615 --> 28:12.115
At the end of my journey to work, I then also have to drive

28:12.625 --> 28:14.515
into the multi-story car park,

28:14.575 --> 28:16.155
at which point I leave the road,

28:16.375 --> 28:18.275
but the car might still be driving,

28:18.415 --> 28:20.555
but it's not driving on the road

28:20.655 --> 28:22.555
and on the multistory car park,

28:22.555 --> 28:24.435
you go up the ramp on the right hand side,

28:24.815 --> 28:26.315
not driving on the left hand side.

28:26.415 --> 28:28.835
So these rules and the way it engages

28:28.835 --> 28:32.795
and understands the situations it in, it's in it,

28:32.815 --> 28:35.555
it crosses a threshold from public road to private,

28:35.745 --> 28:38.075
from normal to abnormal, from driver

28:38.175 --> 28:40.755
to user in control, a user in charge.

28:41.815 --> 28:43.755
The way that these get navigated in practice

28:43.755 --> 28:46.835
and what happens when those thresholds go wrong,

28:47.035 --> 28:50.555
I think is is is going to be really challenging in practice,

28:51.335 --> 28:54.515
to, to understand how those things work out.

28:55.295 --> 28:58.235
The very last thing to say is that when you are

28:59.305 --> 29:01.115
operating in a, in a context

29:01.285 --> 29:02.955
where you are talking about new,

29:03.395 --> 29:06.235
evolving emergent technologies at a high level

29:06.255 --> 29:09.595
of sophistication, there will always be a degree

29:10.375 --> 29:13.635
to which the regulation of the use of that technology has

29:13.635 --> 29:17.035
to defer to the knowledge base of those who know it best,

29:17.245 --> 29:19.875
which is the industry that produces the vehicles.

29:20.615 --> 29:24.315
One concern, I guess the longer term is about the degree

29:24.315 --> 29:28.715
of influence that the automated vehicle industry might be

29:28.715 --> 29:33.075
able to exert over the drawing of some of these thresholds

29:33.375 --> 29:36.715
and the decisions that are made about the way in which the

29:36.715 --> 29:38.435
technology should be used in different

29:38.435 --> 29:39.835
circumstances in the future.

29:40.575 --> 29:43.235
So there are always challenges about how you deal

29:43.235 --> 29:46.995
with those sort of tensions, about who's responsible

29:46.995 --> 29:48.955
for the technology, who knows it best,

29:48.955 --> 29:51.155
and who influences the deci the direction

29:51.425 --> 29:53.075
that regulation takes in the future.

00:00 - The government at least, is of the view that automated vehicles can reduce road harm on the basis that if we can have truly automated vehicles, self-driving cars, that will remove the main cause of road traffic collisions on the roads, which is the human behind the wheel.

00:19 - so human error, whether, whether that's actually possible going forward, I think is a question we, we don't yet know the answer to. you know, every, every year we hear from the likes of Tesla that a fully automated vehicle is just round the corner, and, and that hasn't yet happened. but that is, that is the idea that that automated vehicles will, have the potential at least to be safer than humans because,

00:50 - they will not be distracted in the way humans will, et cetera. And so that's why, the government tasked, the law commission with, a project, to think about how the law should be changed. both the criminal law and civil law, although both of the authors of this piece, myself and and Paul are criminal lawyers, but how the law would need to be changed in order to, to accommodate automated vehicles on our roads, in order

01:20 - that I think, you know, they could support British industry in particular in, in bringing these, these vehicles onto the roads. so that's the potential. Whether it can come to fruition, is another question. But the idea is that the Law Commission would put into place a regulatory framework that would allow that to happen. The Law Commission, took more than three years on this project, which I think is probably the, the most, well,

01:51 - the longest period that they've spent on, on that kind of project. And it was particularly interesting, as I say, because it covered, different areas of law. Usually the law commission is looking at one area of, of the law, whether it be criminal law, whether it be family law, and, and they have team specific to those areas of law. So this was a bit unusual because it was bringing together all sorts of,

02:14 - areas of law. Not only that, but it was a joint project with the Scottish Law Commission. So, you know, we are particularly interested in what the Law commission was saying, in England and Wales. that's jurisdiction in which we work, but this is a joint report with the Scottish Law Commission as well. And so they published the outcome of

02:33 - that report in January, 2022. and given the number of consultation papers, there had been, unsurprisingly, it was a very lengthy report, so over 300 pages now for anyone operating within the law, that's, that's a very large document to have to get your head round. and so I thought it would be useful to, to, to try and, bring out for criminal lawyers in particular, the most important aspects of, of that report. And I asked Paul to join me because

03:04 - although I've got expertise in road traffic law and, and other criminal offenses, he has much more expertise in, in, regulatory frameworks and, and the liability of companies in particular, which is particularly relevant to, to what the law commission had said. Part of what's important about it as a, a topic area is that thinking about self-driving cars kind of brings into question a lot of the fundamental assumptions

03:35 - that tend to underpin thinking in particular about criminal responsibility for driving. You end up having to kind of reset the, the, the, the rules and the frames of reference for how and when you think about questions of liability. So the report itself is talking about automated vehicles, and in particular an automated vehicle is one that, and the term that they, the, they're talking about is it is capable

04:07 - of performing the dynamic driving task by itself. So this is different than any kind of, you know, lane assist or driver assist software that might already exist or be in, in use. an automated vehicle is a, a vehicle, a car that drives itself independent of the driver. And, so in thinking about how they define the, who's responsible for what in this field, the two terms that they talked about, the two actors that they started to

04:45 - think about responsibility for were, if we, if they're not driven by a driver, then who is the person in the car who's in charge of the car or responsible for the car? The term they use for that is a user in charge. So a user in charge is not a driver, they're not doing the driving task, and they are not necessarily meant to be paying attention to the driving of the vehicle while it's driving, but they are supposed to be in a position to respond

05:18 - to an alert or a request or a need to take control of the vehicle. And obviously as the person in charge of the vehicle, the user in charge, they're also responsible for framework issues around the vehicle, like its road worthiness and, and so on and so forth. And the other actor who's really important in the field are what they call the a SD or automated self-driving entity. So this is the operating company that will be providing the software

05:48 - and hardware that allows the vehicle to drive itself and, and running it through their software models, through their support systems and infrastructure. So traditionally, when we think about responsibility for road traffic and accidents on the road, or, you know, deaths, collisions, anything else like that that sits with the driver, unless you are talking about something much further that back down the food chain of, of a manufacturing fault or something similar in this situation,

06:24 - the driver is not driving the car and they don't bear the responsibilities we typically give to a driver as a user in charge. Their responsibilities are much more narrowly defined and framed and much more responsibility for the operation sits with the A SDE, the operator, the the organizational actor that provides the vehicle. So they're having to redevelop new ways of thinking about when those two new kinds of actor become responsible that weren't there before.

06:57 - And that's what was really tricky and challenging for the law commission to do The user in charge, is somebody, who is not expected when the, automated driving system is in control of the, the, the car. So when the, the vehicle itself is, is, engaged in the dynamic driving task, the user in charge isn't expected to do anything in particular. They can pretty much do what they like. Okay. But, as soon as there is that transition demand,

07:31 - once that the, the period comes to an end, so the suggestion is that might be 10 seconds between when the vehicle says you need to take back control of the vehicle. once that happens, that user in charge then kind of metamorphosis in back into a driver. So once they are a driver, that same person who was the user in charge is then liable for any driving offenses that they might commit. And one of the things that the,

07:59 - the law commission were having to think about was, well, what happens if, if somebody who is in the vehicle with a responsibility of the user in charge fails to take back responsibility should recreate a new offense that would, make them liable for failing to do that? And they decided that, that they weren't going to do that, that actually because they then became a driver, they would just be liable for whatever driving offenses they then, committed at that period in time.

08:25 - So if they haven't taken back, control, then presumably, they will be falling below the standard of a competent and careful driver. And so would be committing automatically, if you like, the offensive care that's driving a potentially dangerous driving depending on the circumstances. So that was one of the things they looked at. Another offense in terms of individuals is, is what if, you know, you've got, a group of people there is this vehicle that is capable of, of,

08:54 - being driven by itself, but it does require a user in charge. and none of them want to take that responsibility. So, so a bunch of people get in the car and nobody actually, takes responsibility for sitting in the driving seat. then they have decided that there ought to be, an offense of allowing oneself to be, carried in a vehicle without a user in charge where a user in charge is required.

09:24 - And I suppose what we should mention is that the law commission here had to envisage two types of vehicle, to be able to adapt to, to the way in which technology will be developed. So, so in the first instance, they're, they're thinking about vehicles that probably won't do the full, full jour journey itself in terms of the dynamic driving task that, that, you know, a, a driver may well have to begin the journey, then they can hand over control to the automated,

09:53 - driving system. And then at some point that system will want hand back, responsibility. And, and that's where you have a user in charge. but some vehicles they do envisage in future will, be vehicles where no user in charge is required, which would be in some respects fully automated. But in that, in, in those cases, there would need to be somebody, employed by the A SDE. so the company that runs the vehicle who,

10:22 - who would then, have oversight of the vehicle. So it's, that's not the same as remote driving, but, but they would have some control of the vehicle from a distance. And those, those vehicles, would be ones, where that offense of, of their failing to be a user in charge would, would not be committed because if you don't need a user in charge, then people could just get in the car and, and, carry on.

10:47 - So, that, that's one of the things, I mean, it, it, this is part of the issue here. It's so complex that the, the having to envisage what might be needed and also then the confusion that that can create for human beings. So, so this question of, you know, how, how would somebody know when they got in a vehicle, whether it was one that required a user in charge or not, do they know that,

11:11 - they might be committing an offense if, if there isn't a user in charge and they just let the, the vehicle drive itself? but the benefit of a no user in charge vehicle might be in future that you could have people who would otherwise be incapable of driving perhaps, for, for reasons of disability. and, and, and I think that's one thing as well as the,

11:33 - the road safety aspect. there's this idea that we do want to be able to provide for other members of society who can't at the moment, rely on, on motor vehicles for, for, for transport in the same way as they, they might be able to if there truly was automated vehicles. One of the things that The proposals do is shift a lot of onus away from driver responsibility because you don't have a driver in the way

12:00 - that we traditionally or currently conceive of it. So the question then becomes, where does a lot of that liability go to? And the answer tends to be the operating organizations, the A SDE, the self-driving entity organization that is running or utilizing, providing the technology. And there are really three kinds of offenses that the law commission had envisaged relating to these, automated vehicles. The first ones are kind of around the sale and manufacture

12:32 - and, and advertising of these vehicles that they should only claim to do what they actually do. They shouldn't advertise themselves as self-driving if they're not self-driving and so on. So these are kind of a bit like consumer protection type of offenses. It probably would be fair to say that I think the, automated vehicle industry has not necessarily always observed these principles, up to now in terms of, the way that it sells or,

13:01 - or advertises, quasi autonomous vehicles up to this point. So that kind of motivated the law commission to feel like they needed offenses to draw a bright line distinction between truly self-driving cars and cars that are not truly self-driving, but that will take over crew cruise control, lane assistance, you know, all of these other kind of adjustments that that can help the driver.

13:29 - The second kind of offenses are ones about, I guess kind of regulatory, openness and cooperation. The law commissioner proposed introducing two new regulatory bodies, one of which would be about the manufacture, the safety standards, the provision of automated vehicles. The second one would be a kind of in use regulator to deal with the way that they operate day-to-day on the street, and if there are problems associated with them tending to, you know, not respond to particular hazards or issues

14:08 - or known problems that come up or investigating accidents when they occur. And the law commission introduced new offenses relating to a kind of duty of candor on the companies to share information with those regulators, to provide them details, to allow them to see, everything they need to know about the code and the operating system, and to share details crucially of accidents, or collisions when they occur.

14:37 - again, probably something that, it, it's fair to say, the, emerging automated vehicle industry hasn't always necessarily, been, it was super keen to do in practice, in every eventuality, the most important areas are those related to road traffic infractions, because this is where they're plugging the gap where a driver would normally have been responsible, and they kind of

15:06 - have a model which is based upon the principle of learning from adverse events. So the idea should be that the, this duty to share information is there, self-driving companies have to provide information about a, a collision or an incident or an event. And then the regulator's primary interest would be in using that information to iron out the problems that have led to that issue. Is it about recognizing bus lanes?

15:37 - Is it about being able to identify the difference between, a, you know, an object and a push chair that things that in practice, you know, could potentially be serious problems you want to learn when things go wrong, make adjustments and make collect collections. So it moves away very much from the kind of model that we'd tend to be used, used to of getting a speeding ticket or a, a a a fine or something for, exceeding a speed limit

16:09 - or being in a bus lane or, or whatever else it might be. And taking that approach in relation to those kind of, what you might call factual infractions is probably not that controversial. An issue where it's more controversial probably is in relation to incidents that lead to more serious consequences. So where, a a, a death has been caused, the existing offense is causing death by dangerous driving, causing death by carless driving.

16:42 - They wouldn't map across to these organizational entities because they're not doing the driving, the car is driving itself, so they haven't caused the death by dangerous driving. And by making, that kind of emphasis on learning from those adverse events, the likelihood might be that you end up with a sort of gap where some, a car doesn't do something it's supposed to do, there's a problem that should have been picked up that perhaps that wasn't somebody been

17:17 - plowed into and killed. And the primary approach of the, the model is liable to be about, well, what can we learn from this and how do we put it right for the future? There tend to be a lot of public interest and obviously interest from victims, families and and relatives in, in a, in securing some kind of redress, some kind of accountability, in those situations because, we would be talking about probably higher levels of,

17:46 - of culpability associated with what had happened rather than just, well, it was supposed not to go into that lane and it did go into that lane and now somebody's dead, dear, we'll learn from it. So there's a gap there in relation to potential liability when things go very seriously wrong in that kind of way. But there might be some areas where existing legal measures do plug some of that gap.

18:10 - one example of that, for instance, would be there is a criminal offense of corporate manslaughter. It applies to any corporate organization that, causes a death that which results from a management failure in the way that it organizes or runs its business. And this will apply to any organization, whether it's a building firm or whether it's an automated self-driving entity, operating self-driving cars.

18:37 - So if you could show that management failures in the way that these organizations ran their business and managed the cars that they provided, the vehicles that were out there on the road, it seems that there probably would potentially be some avenue for corporate criminal liability. Now, in practice, in any regulatory setting, those kind of offenses tend to be used very rarely as a last resort and only in the most egregious kind of cases. So it's unlikely that we would see kind of vast numbers

19:11 - of these, of these cases coming before court. But primarily the, the responsibility for, and the accountability for driving offenses as we tend to think about them, it shifted much more to this regulatory learning regulatory pedagogy model than it is the traditional kind of road traffic policing model that we've seen up till now. And that's a trade off that I think the law commission said you kind of have to make, if you want to introduce, a, a reliable

19:45 - high, a reliable system in a, in a high risk novel technological setting. It's the same sort of approach, for instance, that that has traditionally been used in civil aviation, industry learn from the mistakes. But as I said, there are trade offs in accepting that that's the way you wanna go. We have a number of conclusions, and, and I think even since we published the paper, things have obviously developed.

20:11 - and I think, you know, a lot of what we've said still remains true. we, we've just said that, what one of the trade-offs for, for, for what was proposed, is that, those who become bereaved by instance on the roads won't necessarily, get what they might be looking for in terms of criminal liability of either an individual or, or, or a company. And I think, that is, that is gonna be a big challenge

20:46 - that, that that government and, and, and policy makers will have to face is, is responding to the needs of bereaved families. I mean, it's already the case that under the present system, it's felt by many that, that the, the, the needs of bereaved aren't met by the criminal justice system. And, and, and I think managing expectations there is, is going to be really important. but it's this, kind of conceptual leap from, from thinking about things a, a as we have done a as humans

21:19 - to, you know, judging humans behind the wheel of a vehicle that is potentially very dangerous, to handing over responsibility to these, asds, and, and having to trust in to some extent that the technology can do a better job than human beings. but I think, I, one of the, the disappointments I have, coming outta that and one of the recent developments is we, we, one of our conclusions is, the point that the law commission was unable to make the decision as

21:55 - to how safe a vehicle would have to be before it was given authorization to, to, to be on the roads, as, an automated driving system. And, they kind of said, well, that's for government to decide. and they looked at a number of options. Well, of course we now are in a position where government, has put before parliament an act, which actually puts into practice,

22:25 - puts on on the statute book, most of the recommendations of the law commission, and goes further in the sense that it does answer that question as well. So under section two of the automated vehicles at, 2024, which is about the statements of safety principles, it makes the point that, to be authorized a vehicle would have to be as safe, if not safer, than a competent and careful driver. and I think there's a problem there as,

22:55 - as we comment in the article, that that, that kind of perpetuates this idea that we can humanize, these, these robots if you like, which it isn't, helping I think with, with, with that conceptual leap that that, that we as society need to make, that the point is that these, these don't learn like human beings. They're not gonna be distracted like human beings, but, but they, they learn as a fleet. Machine learning is about each, element of a, a fleet of vehicles being able to learn from all the others

23:29 - that are on the roads. and, and I, I personally don't feel that that that test of it being as safe as a competent and careful driver works in any sense of the word. I don't know how, the, the decision maker, will ever decide whether a vehicle is as safe as a competent and careful driver, which leaves open, you know, the, the other question is, is we now have this statute, which puts into practice many of the recommendations, but it's still reliant on a lot of secondary legislation

24:03 - that is yet to be brought in that would create the necessary regulatory bodies that would make these decisions. And so there's still a lot of work to be done there, I think, in, in, in that regard. so I think those, those are some of the main conclusions. And I think the other one is that, you know, these things will take time. And, and my concern from the perspective of somebody who has

24:30 - researched fatalities on the roads and, and how we can try and reduce road date danger in the, long run is that, that there is going to be this transition period, not, not the 10 seconds of handing over, control of a vehicle back to a human being in a particular instance, but a transition period over years, if not decades, where you will have on the roads, both ordinary, what we consider ordinary vehicles where you have a driver, a human behind the wheel who might be making mistakes,

25:03 - but at the same time, you will also have these automated vehicles on the roads, engaging in machine learning, and which may or may not be handing back control to, to a user in charge in the vehicle. And I think, you know, from my perspective in terms of, the campaigns that I see about trying to, to improve road safety or reduce danger on the roads, it's mainly about focusing on the driver and making sure that they are fit

25:33 - to be a driver and am I, you know, the, the worry that, the skills of drivers are deteriorating, the suggestion that we should have graduated driving licenses, et cetera, et cetera. And that kind of all falls by the wayside if we're then, focusing instead on handing everything over to an a computer, if you like, will human beings retain the skill that they have, that is needed to drive vehicles and to be, a user in charge who might be given the responsibility of, of,

26:05 - of driving the fu driving the vehicle at any particular time? I think One of the interesting things and the most challenging things within the, the, the reform proposals and then the legislation that has followed and the implications it has are all about, I guess the concept of thresholds and when thresholds get crossed. So there's a threshold between being a user in charge

26:29 - of a vehicle and then becoming a driver, and how, you know, how much time you are given, how fair it is in those circumstances to expect somebody to suddenly become the driver in charge and drive to the standard of a, a, a competent and careful driver. If the car's already going off a cliff at the point that control is handed back, there's not much you can do as a driver. But there are also these thresholds

26:55 - that we encounter all the time, and automated vehicles have to, we'll have to deal with those. So, you know, as an example on my commute into work, I kind of come past the rugby ground and if there's a big crowd going in, there might be pedestrians in the road waving flags on their way, the car has to adjust to those. But if it can't, if it sees something it doesn't expect or doesn't know, it will either have to respond

27:21 - by giving control to me or it will do what automated vehicles, I think primarily tend to do when something goes wrong or when they try to hand back control and they can't. Or there's a system failure, which is that they will stop themselves. They stop when it's not safe to continue. So they have to deal with stuff they're not expecting to see. And I think the phrase we use in the,

27:45 - or in the, in the paper is about driving and the roads being an open system, not a closed one. It's not self-contained, like, you know, maybe the railways or a monorail would be, everything's on the tracks and it's all contained in that system, and you can automate and control it in the real world. You have stuff coming in left, right, and center permeating it, that that has to be accounted for. At the end of my journey to work, I then also have to drive

28:12 - into the multi-story car park, at which point I leave the road, but the car might still be driving, but it's not driving on the road and on the multistory car park, you go up the ramp on the right hand side, not driving on the left hand side. So these rules and the way it engages and understands the situations it in, it's in it, it crosses a threshold from public road to private,

28:35 - from normal to abnormal, from driver to user in control, a user in charge. The way that these get navigated in practice and what happens when those thresholds go wrong, I think is is is going to be really challenging in practice, to, to understand how those things work out. The very last thing to say is that when you are operating in a, in a context where you are talking about new, evolving emergent technologies at a high level

29:06 - of sophistication, there will always be a degree to which the regulation of the use of that technology has to defer to the knowledge base of those who know it best, which is the industry that produces the vehicles. One concern, I guess the longer term is about the degree of influence that the automated vehicle industry might be able to exert over the drawing of some of these thresholds and the decisions that are made about the way in which the technology should be used in different circumstances in the future.

29:40 - So there are always challenges about how you deal with those sort of tensions, about who's responsible for the technology, who knows it best, and who influences the deci the direction that regulation takes in the future.

Faculti AI can make mistakes