Latest / The Road to Accountable AI / Phil Dawson, Armilla AI: Insurance for AI Risks

Transcript

Phil Dawson, Armilla AI: Insurance for AI Risks · The Road to Accountable AI

0:09
Kevin: Hi, I'm Kevin Warbeck, professor of Legal Studies and
0:11
Business Ethics at the Wharton School of the University of Pennsylvania.
0:15
For decades, I've studied emerging technologies from broadband to blockchain.
0:19
Today, AI is promising to transform our world, but AI needs accountability
0:26
mechanisms to ensure it's developed and deployed in responsible,
0:29
safe, and trustworthy ways.
0:32
On this podcast, I speak
0:33
Dawson: with the experts leading the charge for accountable ai.
0:40
Kevin: Phil Dawson is head of AI Policy and Partnerships at Arm Miller ai.
0:44
We talk about the need and demand for AI specific insurance, what it takes
0:50
to develop effective AI policies, how AI evaluation and assurance mechanisms.
0:55
Feed into insurance underwriting, what policymakers can do to facilitate
1:00
a healthy AI insurance market, what we can learn from cyber insurance to
1:04
apply to the AI context, as well as how AI insurance could create pressure
1:11
for companies to invest effectively and efficiently in AI governance.
1:16
Phil, welcome to the Road to Accountable ai.
1:20
Dawson: Thanks for having me.
1:22
Kevin: Let's start by talking about what exactly AI insurance
1:26
means in specific terms.
1:28
What exactly is that that companies are insuring against?
1:33
Who's the one getting insurance policies?
1:34
How does that process work?
1:36
Dawson: so AI insurance today it's a new category of insurance, but
1:41
it is still fairly heterogeneous.
1:43
So, people think of, AI insurance, when they hear the word as potentially being
1:48
a standalone product such as our Millers.
1:51
But there is an expectation in the market that other policies
1:54
that make no reference whatsoever to AI are also covering AI risk.
2:00
Right?
2:01
So we have, that has kind of been the historical status quo.
2:04
And now when we talk about AI insurance, typically we're meeting one of these,
2:09
standalone products like Arm Miller has standalone, insurance product that
2:13
deals with AI risk and related claims.
2:17
Firstly, that would mean for, let's talk about the insured
2:20
asset as you, as you mentioned.
2:22
For us at Mila, how we think about the insured asset that is the AI
2:26
system or model or system of models.
2:29
So effectively the AI powered application, we're not talking about,
2:33
necessarily the foundation model itself.
2:35
Out of the box chat, GBT, off the shelf we're talking about.
2:40
A model could be an LLM, could be a classical machine learning model,
2:43
or a system of them that have been orchestrated and maybe fine tuned and
2:49
with associated guardrails and used for a specific purpose, or series of purposes.
2:55
And the surface area for those uses is expanding all the time, but
2:59
we're talking about the AI system.
3:01
That is fundamentally driving or powering an application, a software application.
3:06
So the insured asset is an AI product, an AI powered product.
3:09
that's kind of where we start.
3:11
That is what we're insuring.
3:12
And in terms of covered losses, there's two, types of losses, first party losses
3:18
that someone, for instance, using an AI application might suffer if it.
3:22
Hallucinates or leads to model error or some other type of, unintended
3:27
behavior, or harmful output that causes.
3:30
Property damage, for example, to the insured.
3:33
Say it's using ai, in an industrial plant or manufacturing plant, for
3:38
predictive maintenance of equipment.
3:41
And some failure leads to property damage that would be a first party, harm or loss.
3:46
There are also third party harms.
3:48
I think that's where we've seen, the greatest interest
3:51
so far in protection against.
3:53
Third party losses, and that would be where AI used by a company leads
3:58
to some, third party claim against 'em from a customer, a partner, an
4:03
individual who, has suffered themselves some type of harm on account of the ai.
4:09
Underperforming or some error or the systemic discrimination.
4:14
And maybe some of those, class actions that you see in both the health
4:18
insurance base or in AI and hiring and some of the HR tech platforms that
4:22
are being, sued with alleged, bias or systemic bias, that would be a great and
4:26
prominent example of third party harms.
4:29
Kevin: Alright, well let's get into that question as you say, there are lots of
4:32
existing policies that companies have.
4:35
We know whether it's general liability or directors and officers or something else.
4:38
Why do we need a company like Arm Miller to specialize in AI insurance?
4:44
Dawson: There's a really good and also nuanced answer to that question.
4:47
I think why do we need AI insurance policies today?
4:50
Why do we need a company like Armella to help, to launch
4:53
a new category of insurance?
4:56
The most basic way to answer that is to say that traditional insurance policies
5:01
like you mentioned today's directors and office officers, insurance policies,
5:06
errors and admissions policies, tech and o, professional liability cyber
5:09
insurance, they were not designed in their wording or even in their underwriting
5:14
intent to cover AI risk today.
5:18
There was two things there.
5:20
There's not the explicit wording.
5:22
And as part of the policy and the coverage.
5:26
So there's not that clarity in the market, that the market is actually
5:29
really craving right now, given the extent of AI risk and liability exposures.
5:34
So there's not that clarity in the wording.
5:37
But there's also has not as of yet been the intent on the underwriting side.
5:41
So many of the policies today that.
5:44
Make silently cover or address AI risk and do not explicitly exclude it.
5:49
We're not underwritten to the risks.
5:51
As part of a, insurance underwriting process, when an applicant, a large
5:55
company, for instance, applies for one of these insurance products.
5:59
They will typically undergo some type of an assessment,
6:01
let's say, on cyber insurance to understand their risk posture.
6:05
This is changing, but up until recently, mostly these policies and the underwriting
6:10
questions did not make any mention of ai.
6:14
Ai, the applications that are being used, the inventory applications that
6:17
are being used, the specific risks of each of these systems, the policies or
6:21
governance processes that companies have in place to manage and govern AI risk, the
6:26
testing, of these systems, the validation, whether there's third party audits
6:31
or certifications that are obtained.
6:33
This has not been part of the process of.
6:35
Underwriting, in cyber insurance or tech e and o, this is changing and we're starting
6:40
to see questionnaires where insurers are asking more questions about ai.
6:44
but historically this has not been the case.
6:47
When an insured knows that the policy has no wording on ai, and in discussion
6:51
with the broker and potentially also the carrier, they have an understanding that.
6:56
The underwriting process doesn't really address AI risk.
6:59
And there is the assumption then that the pricing has not reflected
7:02
a quantification of AI risk and that at some point the policy could change
7:08
or certain exclusions, for instance.
7:11
Might be relied upon to deny claims if they're in some way, applied to a
7:16
certain scenario or that more practically speaking, there could already be
7:20
gaps, there could already be partial coverage or areas that insurers,
7:26
have a desire to move away from covering.
7:29
And so given the importance of AI in the economy today, given how, important it's
7:35
to some company's growth, particularly if they're software developers or if
7:40
they're investing a lot in AI as a large enterprise, large deployer with
7:45
dozens, potentially hundreds, sometimes of AI uses in production, in material.
7:51
Risk and sensitive uses of highly regulated industries and so on.
7:56
you have very low tolerance for uncertainty in your insurance coverage
8:01
that is supposed to be covering ai.
8:02
You want to know that it covers AI risk.
8:04
You want to see how it covers AI risk.
8:06
You want to understand what limits will be available to you.
8:09
If there is an AI related, loss that leads to a claim, a sizable claim.
8:14
You would prefer today, potentially to pay an insur, an additional insurance
8:19
premium, to have that, security, around the dedicated, AI insurance limits that
8:25
are, that will be available to you.
8:27
And there's an interesting survey from the Geneva Association, which
8:32
is A think tank that is led by 40 of the largest insurers in the world.
8:37
They put out a report this past fall, and a survey of 600.
8:42
Corporate insurance buyers at top, fortune 1000.
8:45
and over 90% of them indicated they're now seeking dedicated
8:49
coverage for ai, generative ai.
8:51
the market is asking for this, and the policy language and the
8:56
underwriting is struggling to keep up.
9:00
Kevin: What then exactly.
9:01
Does ARM do?
9:02
Presumably the writing language to be explicit about AI is the easy part.
9:07
is it that underwriting, that risk is set and help us understand
9:10
where you fit into this landscape?
9:13
Dawson: Yeah.
9:13
So Arm is what you would call a, a managing general agent.
9:16
and we are also a Lloyd's cover, cover Holter.
9:19
So what that means is, is we as a startup insurer.
9:23
We sit between the insurance capital providers or capacity
9:27
providers, as we call them.
9:28
and for us that means, large reinsurers and some insurers who have syndicates
9:32
at Lloyd's who are also interested in backing, specialty insurers who are
9:37
often MGAs, managing general agents who can help, provide, specific expertise,
9:42
technical expertise in areas that are more difficult to underwrite, or new products.
9:47
So, yeah, so we're involved across the board in the drafting of the
9:51
policy wording and our products.
9:52
we've done that with great collaboration from our insurance
9:55
partners, our capacity providers.
9:57
We're also technical underwriters.
9:58
We perform the underwriting at our Miller, and we bind the
10:01
policy as an MG and cover holder.
10:03
when an insured or a broker brings us to one of their customers.
10:07
They deal exclusively with arm and they get an Arm Miller Insurance policy.
10:12
They go through an ARM led underwriting process and Arm Miller binds the
10:15
policy, and access the insurer.
10:19
Kevin: the company didn't start with this business model.
10:22
So, tell us a little bit about the history about how Arm Miller got, to AI insurance.
10:26
Dawson: Arm was founded almost five years ago now.
10:29
I joined the company four years ago.
10:31
When the company was founded, I think the mission has stayed the same, but
10:36
the business has evolved, in search and support of that mission really
10:39
to build justified or evidence-based trust in artificial intelligence.
10:44
and ultimately help with adoption.
10:45
So the company, when it was founded was a SaaS company, that was providing.
10:50
Testing and validation tools to enterprise data science teams as a SaaS platform.
10:55
and this is, five years ago.
10:57
the goal being to provide as companies, particularly larger companies
11:00
who are already expanding in the number of, AI development projects
11:04
that they were embarking upon.
11:05
The idea was that they would need a standardized testing toolkit to be able
11:10
to, for efficiency reasons and then also to more comprehensively have access to the
11:15
most comprehensive toolkit for testing.
11:18
So that was the original company business model.
11:21
But very quickly, after launching the company.
11:24
we saw an opportunity in the insurance market having discussions with reinsurers
11:29
such as Swiss Re, who were already looking into, the potential challenges
11:34
emanating from silent AI coverage, or at least the AI coverage that was there
11:39
in insurance policies, but silent, silently addressed, they were looking at
11:44
this already and trying to understand.
11:46
And they've written some very helpful reports on this over the last several
11:50
years, trying to understand what the impacts could be for the industry
11:54
itself and potentially for insured to the extent that there were gaps in
11:57
coverage or the need for new products to address new market, requirements for ai.
12:04
We thought that this would be, a potentially enormous opportunity.
12:09
and had established some excellent partnerships, including with Swiss
12:12
Re and other partners, Greenlight Re, others who were still working with today.
12:16
And the first product that we launched was a performance guarantee.
12:20
addressing the problem of not only a need for, third party validation
12:25
of AI model performance metrics.
12:29
but more than that, some coverage in the event that they underperformed
12:33
with respect to the KPIs that are often stipulated in the
12:38
service level agreement, the SLA.
12:40
that was our first product and even all the while we were working towards,
12:44
launching this liability insurance product that came out last May.
12:49
as I said, the mission, to build trust in ai, has remained the
12:52
business itself has evolved.
12:54
we performed, dozens of, audits and, and, assessments of applications
12:59
and, and highly regulated contexts and HR and financial services and
13:03
insurance, customer service chatbots.
13:06
all of this helped us build up.
13:09
the scalable evaluation capabilities that we need for underwriting, as well as,
13:16
our own proprietary AI risk data on, the understanding we developed and the data
13:21
on how models fail, how robust, and the robustness requirements of AI systems.
13:27
The probability of failure in different scenarios.
13:31
And yet we notice that still companies, after we had.
13:35
Provided them reports on
13:37
How well models performed or potentially underperformed, they
13:40
became fundamentally more concerned about the risk and exposure of these
13:44
systems, understood the limitations, and the bounds of the performance.
13:49
So we just saw insurance increasingly as we embarked on
13:52
this journey, we saw insurance as.
13:54
a catalyst for trust as a financial incentive for continuous improvement in
13:59
safety and better testing practices, it became more and more clear that insurance
14:05
would provide, a new incentive for companies to invest in the practices and
14:10
the tooling and really the requirements to build and deploy safe, responsible ai.
14:15
Kevin: But for insurance to be that catalyst for trust,
14:19
Dawson: Hmm.
14:19
Kevin: There has to be enough clarity about causation.
14:24
There has to be enough information out there both for the, the insured
14:28
and for, company like Arm Miller.
14:30
and there needs to be some understanding about what are the kinds of practices
14:35
that a firm can and should be doing, and how they might affect the risk.
14:40
Are you able to.
14:41
Address, all of those issues in the current environment.
14:46
Dawson: yes and no, and I think most many of them are definitely works in process.
14:50
you mentioned, the idea of, causation and in some ways that also brings in.
14:55
lack of clarity or poor clarity on liability regimes related to ai.
14:59
yes, I think area that, with more clarification, of the liability
15:03
regime, for instance, or if there was a clear standard by which
15:06
a responsible conduct, could be measured against that would, bring a
15:10
lot of clarity to us as an insurer.
15:13
but for the time being, it's something that we're having to look at on both
15:16
sides is both parties, are concerned about liability, unclear about
15:20
necessarily where it lies and interested in exploring, insurance coverage.
15:25
Kevin: Your job specifically at Arm Miller is head of policy.
15:27
what exactly does that entail and why does an insurance company need
15:31
someone focused on policy issues?
15:34
Dawson: my title is Head of AI policy.
15:36
partnerships in some ways a head of policy title is a little bit of a
15:40
legacy title from our Miller's origins.
15:43
And when we were, launching our SaaS platform and providing
15:47
third party assurance.
15:49
there was always a need for someone to continue to follow shape, participate
15:56
in the public policy conversation on AI governance, and ultimately, nurture
16:02
expertise on AI regulations, standards, ai, risk management frameworks.
16:07
and that has been my career for almost 10 years.
16:10
my first roles in AI policy were 20 17, 20 18, at companies that were
16:15
helping To draft the OECD AI principles.
16:17
I have been part of the AI policy, world for a number of years, and it
16:21
became directly relevant to Arm Miller from a commercial standpoint as we were
16:25
providing the tools to operationalize these policies and these risk
16:30
management and governance frameworks.
16:32
that was still true when we became a, provider of AI assurance services
16:37
and assessments and evaluations.
16:39
companies, our clients.
16:41
very interested in how we had constructed our assessments, based on what
16:45
frameworks, what standards, and we're often looking for guidance, in addition
16:49
to our service on, how they might adjust or develop their own responsible AI
16:54
programs or AI governance programs.
16:56
So there has been kind of a need for Amila to, Showcase leadership, and maintain
17:02
expertise, in ai policy governance, responsible ai, AI safety, and that has
17:07
actually been true in insurance too.
17:09
Kevin: Are there things governments can be doing to facilitate the development
17:15
of a healthy AI insurance market?
17:18
Dawson: Yes, and we get, this is a question we get, I think more and
17:21
more, frequently, including from, state legislatures and lawmakers
17:26
who are interested in better understanding all the ways that.
17:30
public policy or legislation could incentivize better practice beyond
17:37
command and control regulations of prescriptive obligations.
17:40
They want to understand how to activate the market, and tap into market incentives
17:47
for, for self incentivizing behaviors.
17:50
Let's, if we could put it that way.
17:52
So some of the things I think we typically recommend are.
17:57
to explore policy and even legislative measures that could help incentivize
18:04
transparency, disclosure, as well as the emerging AI assurance industry.
18:11
So things that we've already been talking about, that help
18:13
generate the standardized.
18:16
Data that insurance underwriters ultimately need to help
18:21
assess and quantify AI risk.
18:24
and that's because today, one of the challenges of covering ai, as an
18:28
insurer is the absence of claims data
18:31
Even proxies for claims data to build actuarial models and
18:35
underwrite as underwriters would traditionally underwrite.
18:38
and in the absence of that, what you do need is to have access to credible,
18:45
Proxy signals for understanding AI risk.
18:47
and for us, as I mentioned, that is rooted in companies overall,
18:52
governance and risk posture, and critically the AI system itself.
18:56
how performance it is, how robust it is, how safe, how guardrails,
18:59
how performant the guardrails are and to have data in all of these.
19:03
Attributes of a company.
19:05
so if policymakers, as they explore, different policy initiatives or
19:10
even legislation regulation, if they can help incentivize companies to.
19:16
Self-assess to undergo a third party assessment, to, provide,
19:21
explore things like safe harbors to incentivize disclosures and standardize
19:25
what types of disclosures, both in terms of the substance and the
19:28
manner in which they're provided.
19:31
Then we are beginning to have information that is readily
19:34
available for underwriting.
19:36
that is, standard practice in the industry and not a. Special,
19:42
request from a specific underwriter.
19:45
as an example.
19:46
So, if it becomes standard practice to both generate this data and to
19:51
share it commonly with protections and incentives in place, in law.
19:56
Then that can significantly enhance underwriters, ability to access the
20:00
data they require, to underwrite.
20:03
Kevin: does that work given how fast AI is moving?
20:06
you mentioned when you started it was predictive machine learning was the
20:11
primary thing companies are doing.
20:12
Then generative AI and foundation models came along.
20:14
Now so many firms are deploying agentic systems, which have
20:18
a different risk profile.
20:20
So even if companies have to disclose, given the rapid speed of change, is it
20:26
really possible to keep up and underwrite what they're actually doing today?
20:30
Dawson: so that's one of the core challenges.
20:32
one of the many challenges of, both assessing, assuring, evaluating AI
20:37
systems, but also underwriting, them today is just the pace of change, the speed
20:41
at which, companies are experimenting, or relying upon multiple models as part
20:47
of systems and updating them as models are updated and it's extremely dynamic.
20:52
one of the things that Arm Miller does as part of its underwriting,
20:55
and conditions of binding policies, we have certain monitoring.
20:58
Obligations that, an insured would have and other insurers who are looking at
21:03
AI insurance as well are exploring this.
21:05
It's a bit of a new, construct in insurance to have continuing
21:08
duties for policy holders.
21:10
but I think we think it is critical.
21:12
and progressively I think what we would love to see, and we know it's not possible
21:17
for every company today, would be.
21:20
continuous monitoring at greater depths.
21:22
and that's, not possible today and certainly not in place for
21:25
most companies, but there are technologies that enable this.
21:29
there are NL ops platforms and observability platforms
21:32
with evolving capabilities.
21:34
an ability to, both detect, Drift and even get attacks.
21:38
creatively just provide, model output, visibility into model
21:42
outputs on a real time basis.
21:44
I think there in the fullness of time, which is maybe a bit of a
21:49
strange thing to say in the AI industry, but let's just say.
21:52
Looking ahead.
21:54
I think we definitely see the potential to have progressive, monitoring in place.
22:00
and for the time being, depending on the solution for us, it would be
22:04
requesting, updates, when there's material change that has been defined.
22:08
And many companies do define material change in the system,
22:10
and we can also help 'em do that.
22:12
but we would require, Updated information about the system that has undergone a
22:16
material change and certainly annually.
22:18
So that's how we're addressing it today.
22:19
And depending on the use, there may be more, stringent monitoring requirements.
22:24
But the idea is to, be proportionate to really the business reality of
22:29
the customer today while taking into account, our own, needs as an underwriter
22:33
to have sufficient information to underwrite and maintain a policy.
22:38
Kevin: What can we learn from, cyber insurance, which seems like the most
22:42
analogous scenario or, or product.
22:45
There's lots of policies out there, but my understanding is
22:48
the, the perception is that the state of practice in cybersecurity.
22:53
Has not really improved dramatically.
22:55
There hasn't been that virtuous cycle of insurance getting companies
22:59
to improve in the same way.
23:00
are you able to take anything at our Miller from, what has happened in that
23:04
sector to learn how to do better for ai?
23:08
Dawson: Yeah, and it's something that we've, you know, we hear
23:10
many different types of analogies from the cyber insurance space,
23:13
both in terms of how that product.
23:15
emerged and the time it took and some of the exclusions that catalyzed or
23:19
confirmed the category and turned it into, the product that is today
23:23
as well as the evolution of the underwriting and some of the pitfalls.
23:27
some of the challenges of.
23:29
Keeping, staying ahead of different threat vectors and attacks, particularly with
23:34
AI now, part of that, threat landscape.
23:37
so I think there are some good lessons there, for AI insurance as well,
23:42
if it ever is to play that type of incentive for better governance.
23:46
but I think first I'd say is that looking at AI insurance, risk-based pricing.
23:53
Will and should reward organizations with strong
23:56
evaluation and governance practices.
23:58
as in if insurance premiums increase, in proportions of the risk, and if
24:02
risk is measured in part based on how mature companies are from a governance
24:07
risk management standpoint, from a testing, from a control standpoint,
24:11
then the incentive should be aligned.
24:13
That is not always the case with underwriting.
24:14
we do see today in the AI insurance market, different companies
24:19
that are exploring alternative means of underwriting based on
24:22
some of these proxy data sets.
24:23
For instance, AI litigation data, maybe incident data.
24:28
But, I think the consensus is that those data sets are very limited today
24:33
when it comes to underwriting AI risk.
24:35
They can be used as, they're not sufficient proxies.
24:38
mostly it's kind of wild to think, of them as providing the
24:41
foundations for underwriting.
24:42
If you think of litigation data is looking at systems that are being litigated.
24:48
many times those systems are.
24:50
And now we're trying to predict the risks of ag agent systems or MultiGen
24:55
multi-agent systems that are just being developed now, versus, you know, the
25:00
copyright and, and copyright and training data lawsuits that are filing itches bt
25:04
one or two years after it was launched.
25:06
So, it, it, I think there, there are cases today where we see, you see underwriters
25:11
Whose approaches are not aligned with, safety or governance, practices or based
25:17
on them and, and therefore providing incentives for continuous improvement.
25:20
They're, they're looking at, risk, macro level risk signals
25:24
that do not necessarily.
25:26
translate into system specific, incentives for improving
25:30
safety or, or, or risk posture.
25:32
So it can go both ways.
25:35
I think there, we definitely view our approach as helping to incentivize
25:39
evaluations and, incentivize improvements in safety and reduction of risks.
25:44
and in part that's, explained by arm Miller's journey towards, underwriting.
25:48
the second would be that insurers can also require coverage conditions.
25:51
We mentioned duty to monitor, and potentially duty to have a
25:56
third party assessment performed.
25:58
There are different controls that insurer can place or subjectivities
26:01
that insurer can require of an insured as part of binding the policy.
26:04
So there's.
26:06
Quasi, regulatory role that the insurer can play.
26:10
we're ready to buy, but we're gonna need to see these adjustments, over
26:14
the course of the policy period or at renewal, for instance, based on
26:18
the new feature, material change in the application or new deployment.
26:22
I think another thing to take into account is that, premiums
26:25
should, scale with deployment size, at risk and risk exposure.
26:30
some traditional underwriting processes for other insurance
26:32
products might look at.
26:34
Some things like company revenue and that would play a significant role as part of
26:38
the pricing and amongst other factors.
26:41
It's a little bit more challenging for artificial intelligence where you can
26:44
have very small providers, who become ones successful very, very quickly
26:50
or, land their first client, which is a Fortune 500 company that uses
26:55
their hr, ai hiring system at scale.
26:58
so the exposure, the profile, is a little bit different.
27:01
and the premiums can and should take account of that.
27:04
so that as a company grows, or depending on the type of deployment, the type
27:08
of customer it has, the exposure profiles is very, very different.
27:12
And pricing should reflect that.
27:14
So, one thing we've done at Arm Miller is we've launched a partnership program
27:19
Kevin: that,
27:20
Dawson: Well known AI, GRC platforms or auditing companies, certification
27:25
bodies, assessment providers into, services that we can offer to our insured.
27:30
So now we're.
27:32
investing time and support and helping to create, for companies who don't
27:36
have already this whole marketplace of technologies and services that
27:41
can help them improve, their posture.
27:44
of course that makes them more ready to be underwritten by Mila.
27:48
but there's a role there, for insurers in helping to improve the overall
27:51
maturity of the industry, as it begins to explore insurance and insurability.
27:57
and beyond that, think of some of the things that we mentioned before
28:00
investing time and shaping industry standards and some of the research into
28:04
areas around liability, for instance.
28:06
These are things that we're doing at our mill.
28:08
and that other insurers, would naturally have an incentive to do
28:11
as well, to help build the market.
28:14
Kevin: it's a really exciting opportunity, and, fascinating to watch, as you
28:19
and others are trying to build this.
28:20
we have to wrap up.
28:21
And Phil, thanks so much for being part of this podcast.
28:24
Dawson: Thank you so much for asking me to be on the show today.
28:29
Kevin: This has been The Road
28:30
Dawson: If you like what you're hearing, please give us a good
28:32
review Thank you for listening.
28:37
Kevin: If you want to go deeper on AI governance, trust and responsibility
28:41
with me and other distinguished faculty sign up for the next cohort of
28:44
Wharton's Strategies for Accountable AI Online Executive Education program.
28:49
Featuring live interaction with faculty expert interviews and custom
28:53
designed asynchronous content.
28:55
Join fellow business leaders to learn valuable skills you can
28:58
put to work in your organization.
29:00
Visit exec ed.wharton.upen.edu/acai for full details.
29:08
I hope to see you there.