3001: Faculty Why Every Business Must Become AI-Driven

[00:00:03] [SPEAKER_00]: Our business is truly ready for the AI-driven future.

[00:00:09] [SPEAKER_00]: Well today I'm going to be joined by Dr Mark Warner and he is the CEO and co-founder of

[00:00:15] [SPEAKER_00]: a company called Faculty.

[00:00:17] [SPEAKER_00]: And this company is dedicated to deploying safe AI systems that are able to combine human

[00:00:24] [SPEAKER_00]: expertise with the latest cutting edge technology.

[00:00:27] [SPEAKER_00]: And Mark's journey from physicist specializing in quantum sensing at Harvard to leading one

[00:00:34] [SPEAKER_00]: of the UK's most innovative AI firms has equipped him with unparalleled insights into the transformative

[00:00:42] [SPEAKER_00]: power of AI.

[00:00:44] [SPEAKER_00]: So in our conversation today we're going to be talking about why Mark believes over

[00:00:48] [SPEAKER_00]: the next decade every business will need to embrace AI or risk extinction.

[00:00:54] [SPEAKER_00]: Big words.

[00:00:55] [SPEAKER_00]: And we'll also discuss his pragmatic views on AI regulation arguing for a more balanced

[00:01:02] [SPEAKER_00]: approach that fosters innovation while ensuring safety.

[00:01:05] [SPEAKER_00]: And Mark will also share how leaders who proactively engage with AI safety will not only safeguard

[00:01:11] [SPEAKER_00]: their organization but also position themselves at the forefront of this technological revolution.

[00:01:19] [SPEAKER_00]: So what does it take to integrate AI responsibly and effectively and how can you or your

[00:01:24] [SPEAKER_00]: business align your AI strategy with their core objectives to truly thrive in this emerging arena?

[00:01:33] [SPEAKER_00]: I just want to take a time out to express my gratitude to everyone who supports our mission

[00:01:38] [SPEAKER_00]: of delivering content every day to 140,000 listeners across 165 countries.

[00:01:44] [SPEAKER_00]: I'm grateful for the support that allows me to maintain this Daily Tech podcast.

[00:01:48] [SPEAKER_00]: And it's also an opportunity to talk about the fact that legacy managed file transfer tools are

[00:01:53] [SPEAKER_00]: outdated and lack the security that today's remote workforce often demands.

[00:01:58] [SPEAKER_00]: And companies that continue relying on that outdated technology, they can put their sensitive

[00:02:02] [SPEAKER_00]: data at risk. So if security breaches are keeping you up at night, why not sleep

[00:02:07] [SPEAKER_00]: soundly with KeiWorks MFT suite because their software security hardening includes

[00:02:12] [SPEAKER_00]: an ongoing bounty program, regular penetration testing.

[00:02:16] [SPEAKER_00]: And with one click appliance updates, staying secure has actually never been easy.

[00:02:21] [SPEAKER_00]: It doesn't have to be complicated.

[00:02:23] [SPEAKER_00]: So while other solutions might leave you vulnerable, KeiWorks offers military grade

[00:02:27] [SPEAKER_00]: protection, so there's no need to compromise on security.

[00:02:31] [SPEAKER_00]: So step into the future of secure managed file transfer with KeiWorks by visiting

[00:02:35] [SPEAKER_00]: KeiWorks.com to get started.

[00:02:38] [SPEAKER_00]: That's right, KeiWorks.com to get started today.

[00:02:41] [SPEAKER_00]: It's time to shift gears somewhat and to do

[00:02:43] [SPEAKER_00]: the person that you've all been waiting for today.

[00:02:47] [SPEAKER_00]: Well, buckle up and hold on tight as I beam your ears all the way to the UK where

[00:02:52] [SPEAKER_00]: you can join myself and Dr Mark Warner, CEO and co-founder of faculty,

[00:02:57] [SPEAKER_00]: where we'll talk about all this and much more.

[00:03:01] [SPEAKER_00]: So a massive warm welcome to the show, Mark.

[00:03:04] [SPEAKER_00]: Can you tell everyone listening a little bit who you are and what you do?

[00:03:08] [SPEAKER_01]: Well, thank you very much for inviting me on.

[00:03:10] [SPEAKER_01]: So I'm Mark Warner.

[00:03:12] [SPEAKER_01]: I'm the CEO of a company called Faculty.

[00:03:16] [SPEAKER_01]: We're an applied AI company, which means we

[00:03:19] [SPEAKER_01]: hate this very powerful technology that is currently being developed and put it

[00:03:25] [SPEAKER_01]: to use in real world situations.

[00:03:28] [SPEAKER_01]: So trying to make hospitals function better or companies, products cheaper

[00:03:34] [SPEAKER_01]: or government services or efficient for citizens, those kinds of things.

[00:03:40] [SPEAKER_00]: Well, it's a pleasure to have you join me on the podcast today.

[00:03:43] [SPEAKER_00]: You're right in the heart of this space and there's so much hype around AI at the moment.

[00:03:48] [SPEAKER_00]: I was reading a great Bill Gates quote a few days ago and he said,

[00:03:51] [SPEAKER_00]: we overestimate what we can achieve in one year, but underestimate what we can

[00:03:55] [SPEAKER_00]: achieve in 10. So I'm curious from everything that you're seeing,

[00:03:58] [SPEAKER_00]: how do you envision the role of AI evolving throughout businesses over

[00:04:02] [SPEAKER_00]: the next decade? And are there any particular steps that organisations

[00:04:06] [SPEAKER_00]: should actually be preparing for and taking right now to prepare for this transformation?

[00:04:11] [SPEAKER_00]: Because there's so much more to come than just gen AI, isn't it?

[00:04:15] [SPEAKER_01]: Hottily agreed.

[00:04:16] [SPEAKER_01]: So I do think that in some ways AI is over hyped in the short run.

[00:04:23] [SPEAKER_00]: Yeah.

[00:04:23] [SPEAKER_01]: And we are going to see a bunch of companies putting money into AI projects,

[00:04:31] [SPEAKER_01]: relying on people who were experts in crypto a year ago and experts in VR two years

[00:04:37] [SPEAKER_01]: before that, and those aren't going to go that well.

[00:04:41] [SPEAKER_01]: And so I do think there'll be a bit of a backlash against that.

[00:04:45] [SPEAKER_01]: However, all of this is on the sort of foundations of the most important

[00:04:52] [SPEAKER_01]: technological transformation of our era.

[00:04:57] [SPEAKER_01]: It'll be like the Internet.

[00:04:58] [SPEAKER_01]: There'll be, you know, there was a dot-com bubble, but 10, 20 years later,

[00:05:03] [SPEAKER_01]: the world's most valuable companies of all sort of Internet or Internet adjacent

[00:05:10] [SPEAKER_01]: companies. So what's actually the deep reality that underlies all of this?

[00:05:19] [SPEAKER_01]: Well, for 20 years, enterprises focused on digital transformation.

[00:05:24] [SPEAKER_01]: Some did this very successfully.

[00:05:27] [SPEAKER_01]: You know, Netflix went from post-DVDs to being a world leading streamer,

[00:05:31] [SPEAKER_01]: but lots failed.

[00:05:33] [SPEAKER_01]: Only about 10, 12 percent of companies sustained their digital transformation

[00:05:38] [SPEAKER_01]: goals for three plus years.

[00:05:42] [SPEAKER_01]: That happened for a lot of reasons.

[00:05:46] [SPEAKER_01]: I'm merely that

[00:05:50] [SPEAKER_01]: transformation is really a sort of what I call a socio technological problem.

[00:05:55] [SPEAKER_01]: It's not just a technology problem.

[00:05:56] [SPEAKER_01]: There's a real human aspect to it as well.

[00:06:01] [SPEAKER_01]: And now the area has begun over the next decade, every business must become

[00:06:06] [SPEAKER_01]: an AI business to go bust.

[00:06:09] [SPEAKER_01]: And so organizations are going to be faced with this sort of

[00:06:12] [SPEAKER_01]: socio technological problem of how do we build the new technology into our

[00:06:21] [SPEAKER_01]: core business while managing its risks and how do we create the capability

[00:06:26] [SPEAKER_01]: to decide our organization that enables us to go to actually put it to use.

[00:06:33] [SPEAKER_00]: So many great points there and even just looking at the Fortune 500 company,

[00:06:38] [SPEAKER_00]: I think there's something like 52 percent of the Fortune 500 have gone

[00:06:42] [SPEAKER_00]: since the year 2000, which is quite a sobering thought as well.

[00:06:46] [SPEAKER_00]: And one of the things I love doing on this podcast is trying to demystify

[00:06:51] [SPEAKER_00]: some of the phrases that we might hear now again.

[00:06:54] [SPEAKER_00]: And one that I keep keep hearing about is the concept of mundane AI regulation.

[00:06:59] [SPEAKER_00]: Can you tell me a bit more about that and why you think it's crucial for

[00:07:02] [SPEAKER_00]: ensuring the safe deployment of AI systems because again, huge talking point right now.

[00:07:09] [SPEAKER_01]: Absolutely.

[00:07:10] [SPEAKER_01]: So well, firstly, AI is a very broad term.

[00:07:15] [SPEAKER_01]: Nobody says biology is unequivocably, unequivocably good or bad.

[00:07:21] [SPEAKER_01]: And you don't talk about regulating biology.

[00:07:25] [SPEAKER_01]: You talk about regulating drugs or

[00:07:28] [SPEAKER_01]: crops or genetic engineering or IVF.

[00:07:33] [SPEAKER_01]: All of those sit under this broad heading of biology, but

[00:07:37] [SPEAKER_01]: to try and talk about them all at once is incredibly complicated.

[00:07:43] [SPEAKER_01]: I think it's useful in the domain of AI to kind of create this slightly

[00:07:48] [SPEAKER_01]: reimaginary spectrum from narrow AI to general AI.

[00:07:54] [SPEAKER_01]: And then on the narrow end, you've got AI models that have been used in production

[00:08:00] [SPEAKER_01]: for decades, things like handwriting recognition

[00:08:04] [SPEAKER_01]: that are specifically geared to do a specific problem and execute on that very effectively.

[00:08:13] [SPEAKER_01]: And then on the other end, the general end,

[00:08:16] [SPEAKER_01]: that's where the AI is supposed to be a sort of universal problem solver like humans.

[00:08:23] [SPEAKER_01]: Now, we don't currently understand how to build this and we don't currently

[00:08:26] [SPEAKER_01]: understand how to control it.

[00:08:28] [SPEAKER_01]: So at that end of the spectrum, I think we should be very cautious.

[00:08:33] [SPEAKER_01]: But at the mundane end of the spectrum, at the narrow end,

[00:08:38] [SPEAKER_01]: we actually should make sure we regulate effectively, but we should do it on the

[00:08:43] [SPEAKER_01]: basis of the domain that we're that it's being used in.

[00:08:48] [SPEAKER_01]: You know, I think the risks of whether some small

[00:08:54] [SPEAKER_01]: fashion company puts a red t-shirt or a blue t-shirt at the top of their website

[00:08:59] [SPEAKER_01]: are relatively low and the burden that they should be then

[00:09:04] [SPEAKER_01]: kind of regulatory burden that they should feel should be very small.

[00:09:08] [SPEAKER_01]: On the other hand, if an AI model is making sort of cancer diagnosis

[00:09:13] [SPEAKER_01]: or something that really has life or death impact, then

[00:09:16] [SPEAKER_01]: you know, there should be serious thought that goes into how to make that,

[00:09:22] [SPEAKER_01]: how to make that safe.

[00:09:25] [SPEAKER_01]: The other aspect to angle on all of this is that for much of Western Europe,

[00:09:30] [SPEAKER_01]: our productivity has been lightning for about 15 years, roughly since the financial crash.

[00:09:36] [SPEAKER_01]: And so we do need to figure out ways to grow our economies again.

[00:09:41] [SPEAKER_01]: And I think narrow AI offers a really important possibility for doing that.

[00:09:48] [SPEAKER_01]: And so I'd like to see us go fast on narrow AI with sensible domain specific

[00:09:54] [SPEAKER_01]: regulation and be cautious on general AI with more thought going in.

[00:10:04] [SPEAKER_00]: And I'm curious, I've got to ask you this question because you flirted with AGI

[00:10:10] [SPEAKER_00]: a little there is Google's, I think it was DeepMind co-founder Shane Lake.

[00:10:14] [SPEAKER_00]: He said in an interview that there's a 50% chance of AGI can be achieved by 2028.

[00:10:19] [SPEAKER_00]: And then I think Elon Musk predicted 2029.

[00:10:22] [SPEAKER_00]: And all kind of start up founders having bets against or for Musk and his predictions.

[00:10:28] [SPEAKER_00]: And it's even been brought forward in some circles.

[00:10:31] [SPEAKER_00]: Realistically, how far away are we from this?

[00:10:33] [SPEAKER_01]: Do you think the truth is nobody knows?

[00:10:36] [SPEAKER_01]: Yes.

[00:10:38] [SPEAKER_01]: So we still, you know, it what's fueled this recent excitement is the incredible

[00:10:46] [SPEAKER_01]: progress that these large language models and large multimodal models have made.

[00:10:51] [SPEAKER_01]: They've turned out to be more powerful than any of us really anticipated.

[00:10:57] [SPEAKER_01]: And so it's only natural to ask how far will they go?

[00:11:02] [SPEAKER_01]: And.

[00:11:03] [SPEAKER_01]: You know, science and the growth of knowledge is one of those really genuinely

[00:11:10] [SPEAKER_01]: unpredictable things.

[00:11:12] [SPEAKER_01]: And so.

[00:11:15] [SPEAKER_01]: Like there is a great deal of debate in the field.

[00:11:19] [SPEAKER_01]: I guess

[00:11:23] [SPEAKER_01]: Shane and people like him who point to the sort of 2028, 2029.

[00:11:28] [SPEAKER_01]: They're basically trying to judge off the back of Moore's law.

[00:11:34] [SPEAKER_01]: And like when we start to be able to build computers that kind of feel like they

[00:11:40] [SPEAKER_01]: have approximately similar computational power to a human brain.

[00:11:44] [SPEAKER_01]: And so that's that's the basis for those kinds of numbers,

[00:11:48] [SPEAKER_01]: which truthfully is as good a basis as any that we have.

[00:11:52] [SPEAKER_01]: Having said that, it can often take us a bit longer than we

[00:11:58] [SPEAKER_01]: anticipate to actually get things to work.

[00:12:01] [SPEAKER_01]: So I think it's probably going to be slightly slower than 2028, 2029,

[00:12:06] [SPEAKER_01]: but probably not 50 years away, maybe like five or 10 years after that.

[00:12:13] [SPEAKER_00]: And AI is such a vast topic.

[00:12:15] [SPEAKER_00]: AGI is a topic on its own for an entire episode of a podcast.

[00:12:19] [SPEAKER_00]: I'm going to further complicate things here and giving your background in

[00:12:24] [SPEAKER_00]: quantum sensing and academic research.

[00:12:27] [SPEAKER_00]: How do you see these fields intersecting with AI to drive innovation in the near

[00:12:33] [SPEAKER_00]: future, because we're starting to hear more and more about quantum alongside AI?

[00:12:37] [SPEAKER_00]: I'm just curious how you see all this panning up.

[00:12:40] [SPEAKER_01]: Well, I sort of voted with my feet on this.

[00:12:43] [SPEAKER_01]: So I was a quantum physicist, a research fellow at Harvard most recently.

[00:12:52] [SPEAKER_01]: And looking at the two fields,

[00:12:55] [SPEAKER_01]: I decided that I thought AI was going to be fundamentally more impactful.

[00:13:01] [SPEAKER_01]: And that impact was going to come faster than quantum.

[00:13:06] [SPEAKER_01]: So I actually moved fields to about 10 years ago to work in AI.

[00:13:15] [SPEAKER_01]: At the moment, we don't have like great evidence, at least as far as I've

[00:13:24] [SPEAKER_01]: followed the field that we actually are going to get important quantum speedups

[00:13:29] [SPEAKER_01]: for anything AI related, although that remains a possibility.

[00:13:33] [SPEAKER_01]: We definitely can't say that we won't.

[00:13:35] [SPEAKER_01]: But I don't think there's a killer application for quantum computers in AI yet.

[00:13:39] [SPEAKER_00]: And what would you say are some of the most impactful AI solutions faculty has

[00:13:44] [SPEAKER_00]: developed? Because I know you work with so many vast areas from government

[00:13:48] [SPEAKER_00]: agencies to leading brands.

[00:13:51] [SPEAKER_00]: So what have you seen here and what kind of outcomes have they achieved

[00:13:54] [SPEAKER_00]: with working with you?

[00:13:57] [SPEAKER_00]: Because this is where the magic happens, isn't it?

[00:13:58] [SPEAKER_00]: We're solving real problems and making measurable impact.

[00:14:01] [SPEAKER_01]: So I think maybe let's do a couple on the public sector and a couple on the private sector.

[00:14:07] [SPEAKER_01]: One of our early most famous results was that we

[00:14:11] [SPEAKER_01]: proved that we built the technology that proved AI terrorist content

[00:14:17] [SPEAKER_01]: moderation could happen at the web scale.

[00:14:21] [SPEAKER_01]: So this was under Prime Minister Theresa May

[00:14:23] [SPEAKER_01]: and we demonstrated that AI algorithms could detect

[00:14:29] [SPEAKER_01]: terrorist propaganda content with enough accuracy that you could start to

[00:14:34] [SPEAKER_01]: effectively, automatically moderate

[00:14:38] [SPEAKER_01]: internet video platforms, things like YouTube or whatever.

[00:14:42] [SPEAKER_01]: And off the back of that, the Prime Minister went to the UN and actually

[00:14:45] [SPEAKER_01]: demanded that the tech companies take down terrorist propaganda within two hours.

[00:14:50] [SPEAKER_01]: So that was a huge, huge positive for the world, I think,

[00:14:55] [SPEAKER_01]: and actually moved the needle in helping fight terrorism.

[00:15:01] [SPEAKER_01]: Then perhaps more recently, we built the technology that the NHS

[00:15:07] [SPEAKER_01]: used to understand the national pandemic response in the UK,

[00:15:13] [SPEAKER_01]: which has been credited with saving thousands of lives.

[00:15:17] [SPEAKER_01]: And that was predictions for every hospital across the entire country

[00:15:23] [SPEAKER_01]: for how many COVID cases they were likely to face in the next three weeks so that

[00:15:30] [SPEAKER_01]: they could figure out where to move patients or oxygen or ventilators or all

[00:15:35] [SPEAKER_01]: those kinds of things.

[00:15:37] [SPEAKER_01]: So those two really nice, important outcomes on the public sector.

[00:15:42] [SPEAKER_01]: In the private sector,

[00:15:47] [SPEAKER_01]: we've had equivalently important

[00:15:52] [SPEAKER_01]: commercial outcomes for organisations.

[00:15:54] [SPEAKER_01]: So we built technology that roots

[00:15:58] [SPEAKER_01]: millions, tens of millions of customer calls a year to the optimal call centre

[00:16:04] [SPEAKER_01]: agents so that people get to speak with the call centre agent

[00:16:09] [SPEAKER_01]: that is best for answering their problems.

[00:16:13] [SPEAKER_01]: Because I'm sure you like me have had a fair number of experiences of calling

[00:16:18] [SPEAKER_01]: the calling these hotlines and the ability of an agent is not to answer

[00:16:25] [SPEAKER_01]: a particular question can be a little bit lacking, but if you can find

[00:16:29] [SPEAKER_01]: exactly the right agent to help you with exactly the right problem,

[00:16:32] [SPEAKER_01]: flies through it makes the call shorter, which makes it cheaper for the company.

[00:16:38] [SPEAKER_01]: But most importantly, it makes it

[00:16:40] [SPEAKER_01]: better for the customer.

[00:16:43] [SPEAKER_01]: And another nice example.

[00:16:45] [SPEAKER_01]: So one of the world's leading scientific publishers, we help them

[00:16:51] [SPEAKER_01]: help their users to navigate through their scientific content.

[00:16:56] [SPEAKER_01]: So here's the problem I faced when I was a researcher that there are so

[00:17:02] [SPEAKER_01]: many papers out there.

[00:17:05] [SPEAKER_01]: It's hard to know what you should read and it's hard to know what you should

[00:17:10] [SPEAKER_01]: read next if you've just finished reading a paper.

[00:17:12] [SPEAKER_01]: And so we built them some technology

[00:17:17] [SPEAKER_01]: to build these personalized journeys like

[00:17:20] [SPEAKER_01]: personalized learning journeys through their scientific literature.

[00:17:24] [SPEAKER_01]: And that significantly increases the click through rate.

[00:17:27] [SPEAKER_01]: So really shows that users are engaging with the content more

[00:17:30] [SPEAKER_01]: effectively than they were before.

[00:17:33] [SPEAKER_00]: Well, I love about our conversation today is these are real world examples

[00:17:37] [SPEAKER_00]: that are moving the needle right now, not the future, but today.

[00:17:42] [SPEAKER_00]: Someone's food for thought for business leaders and for IT leaders.

[00:17:46] [SPEAKER_00]: They're biggest concerns around the pace of this technology is going to be

[00:17:50] [SPEAKER_00]: safety and responsibility and using AI ethically, etc.

[00:17:54] [SPEAKER_00]: So during the recent, I think it was the AI

[00:17:56] [SPEAKER_00]: safety summit at Bletchley Park.

[00:17:59] [SPEAKER_00]: What were the key takeaways and how do you think they will influence things

[00:18:03] [SPEAKER_00]: like the future of policy and regulation moving forward?

[00:18:07] [SPEAKER_00]: Because so important, it's easy to get distracted by the nice new shiny tech.

[00:18:11] [SPEAKER_00]: And we've seen what happens in the past with when Silicon Valley moves fast

[00:18:15] [SPEAKER_00]: and breaks things. So how do you say this going differently this time?

[00:18:20] [SPEAKER_01]: Well, I think probably the most important element was that just

[00:18:26] [SPEAKER_01]: holding the summit is progress.

[00:18:29] [SPEAKER_01]: So imagine if same energy had been applied to global warming before it became

[00:18:34] [SPEAKER_01]: a serious problem and countries started working to generate

[00:18:39] [SPEAKER_01]: international agreement on how we'd solve it.

[00:18:42] [SPEAKER_01]: We've been in a much better position today.

[00:18:45] [SPEAKER_01]: So I think that in itself is wonderful.

[00:18:50] [SPEAKER_01]: There's sort of the key takeaway is there was the Bletchley declaration

[00:18:54] [SPEAKER_01]: where companies, countries came together, actually companies and countries,

[00:18:58] [SPEAKER_01]: but were part of it.

[00:19:01] [SPEAKER_01]: But it was the countries that signed the declaration

[00:19:03] [SPEAKER_01]: to say that they recognized that AI had a lot to offer,

[00:19:09] [SPEAKER_01]: but it was important to safely, which is important.

[00:19:12] [SPEAKER_01]: Both China and America signing that is a big deal.

[00:19:16] [SPEAKER_01]: And then there was agreement for some of the Frontier Labs or the Frontier Labs

[00:19:21] [SPEAKER_01]: that were there to hand over the details of their models for inspection

[00:19:26] [SPEAKER_01]: pre-release by the government, in particular by the UK's AI Safety Institute,

[00:19:33] [SPEAKER_01]: which is a wonderful thing and really great progress.

[00:19:38] [SPEAKER_01]: Now, I think that it's important over time that the government starts to set

[00:19:44] [SPEAKER_01]: standards that can be more widely

[00:19:51] [SPEAKER_01]: used to test to so that everybody knows where they stand rather than just

[00:19:55] [SPEAKER_01]: a small number of Frontier Labs.

[00:19:58] [SPEAKER_01]: But I also think that it's absolutely sensible to start with the people at

[00:20:05] [SPEAKER_01]: the cutting edge of the field and then over time and with more experience,

[00:20:09] [SPEAKER_01]: roll those standards out more broadly.

[00:20:14] [SPEAKER_00]: And you mentioned climate change a few moments ago there.

[00:20:17] [SPEAKER_00]: And I think more recently, AI and sustainability are often mentioned in the same

[00:20:22] [SPEAKER_00]: sentence, but AI also has this well documented energy problem with the amount

[00:20:26] [SPEAKER_00]: it consumes, whether it be water for water, calling all just the energy to keep

[00:20:30] [SPEAKER_00]: those data centers going.

[00:20:32] [SPEAKER_00]: How do you see that conflict resolving itself?

[00:20:39] [SPEAKER_01]: I mean, I think it is going to

[00:20:45] [SPEAKER_01]: these large models do require large amounts of electricity.

[00:20:48] [SPEAKER_01]: Interesting.

[00:20:48] [SPEAKER_00]: Yeah.

[00:20:50] [SPEAKER_01]: But we do know how to generate that in a renewable way.

[00:20:55] [SPEAKER_01]: And so

[00:20:57] [SPEAKER_01]: for large organizations, particularly, I think it's important that they take

[00:21:03] [SPEAKER_01]: the environmental consideration seriously.

[00:21:06] [SPEAKER_01]: But given the amount of money that they're spending on

[00:21:12] [SPEAKER_01]: the chips for these systems, which is in the like 50 to 100 billion

[00:21:18] [SPEAKER_01]: a year for the tech giants,

[00:21:22] [SPEAKER_01]: actually just making sure they run off for a new electric electricity should be

[00:21:27] [SPEAKER_01]: like kind of easy and something they should just do.

[00:21:31] [SPEAKER_01]: Yeah.

[00:21:33] [SPEAKER_00]: Great point.

[00:21:34] [SPEAKER_00]: And as AI does become more integrated into business operations,

[00:21:38] [SPEAKER_00]: are there any other ethical considerations that leaders should be

[00:21:41] [SPEAKER_00]: prioritizing to ensure responsible and fair use of AI technologies?

[00:21:48] [SPEAKER_01]: Yeah.

[00:21:49] [SPEAKER_01]: Well, this is one of those nice cases when

[00:21:55] [SPEAKER_01]: the effectiveness of the solution and the ethics and safety actually go quite

[00:22:02] [SPEAKER_01]: hand in hand.

[00:22:04] [SPEAKER_01]: So,

[00:22:05] [SPEAKER_01]: you know, at the moment,

[00:22:08] [SPEAKER_01]: people talk about

[00:22:12] [SPEAKER_01]: AI replacing jobs.

[00:22:14] [SPEAKER_01]: Now, we've never seen that happen.

[00:22:16] [SPEAKER_01]: AI is not human level.

[00:22:19] [SPEAKER_01]: And so it's not capable of just wholly replacing aspects of

[00:22:27] [SPEAKER_01]: every job that a person does.

[00:22:31] [SPEAKER_01]: What it can do is take individual tasks

[00:22:35] [SPEAKER_01]: and it can automate a sort of a task.

[00:22:40] [SPEAKER_01]: And that basically lets the human divert their attention away from that generally

[00:22:46] [SPEAKER_01]: fairly mundane task to the more interesting things.

[00:22:50] [SPEAKER_01]: Now, the slight problem at the moment

[00:22:55] [SPEAKER_01]: is the way we're building these AI algorithms is just sort of

[00:23:01] [SPEAKER_01]: or the way many people are building these AI algorithms is

[00:23:05] [SPEAKER_01]: actually kind of broken.

[00:23:07] [SPEAKER_01]: They, well, let me give you an analogy.

[00:23:09] [SPEAKER_01]: Imagine you were hiring some people for your team, Neil.

[00:23:15] [SPEAKER_01]: And I said, well, you can hire,

[00:23:18] [SPEAKER_01]: we're going to hire you some people,

[00:23:22] [SPEAKER_01]: but they can't talk to their colleagues.

[00:23:24] [SPEAKER_01]: They can't talk to their boss and you can't tell them about your company

[00:23:29] [SPEAKER_01]: policies.

[00:23:31] [SPEAKER_01]: How well do you think that's going to work out for you?

[00:23:34] [SPEAKER_00]: I think I'll come crashing down very quickly.

[00:23:36] [SPEAKER_01]: Yeah.

[00:23:37] [SPEAKER_01]: And yet at the moment we're building AI models in a way

[00:23:43] [SPEAKER_01]: that is completely disconnected from other AI models.

[00:23:46] [SPEAKER_01]: So each one is kind of locally optimizing in a silo equivalent

[00:23:51] [SPEAKER_01]: of not talking to their colleagues.

[00:23:53] [SPEAKER_01]: We're making them so that they're

[00:23:55] [SPEAKER_01]: completely disconnected from business users.

[00:23:58] [SPEAKER_01]: You know, only the sort of ivory tower data scientists can really

[00:24:03] [SPEAKER_01]: interact with these AI models.

[00:24:05] [SPEAKER_01]: The actual business users can't.

[00:24:07] [SPEAKER_01]: So it's effectively not being able to talk to their boss again.

[00:24:10] [SPEAKER_01]: Another terrible thing.

[00:24:13] [SPEAKER_01]: And then we don't have very good governance around the AI model.

[00:24:18] [SPEAKER_01]: So you sort of train it and then you deploy it.

[00:24:21] [SPEAKER_01]: And you in most circumstances people then hope that it works.

[00:24:26] [SPEAKER_01]: Again, like, you know, hopefully if you hide in your employee,

[00:24:30] [SPEAKER_01]: you tell them what was and wasn't acceptable according to your company policies.

[00:24:34] [SPEAKER_01]: And you'd make sure those were reasonably strictly enforced.

[00:24:37] [SPEAKER_01]: We just don't do that at the moment.

[00:24:41] [SPEAKER_01]: So if business leaders want to prioritize responsible and fair use of AI

[00:24:47] [SPEAKER_01]: technologies, there's a relatively simple answer.

[00:24:51] [SPEAKER_01]: Which is.

[00:24:53] [SPEAKER_01]: Build it into your organizations to empower people to make better decisions.

[00:25:00] [SPEAKER_01]: If you do that and you ensure that, one,

[00:25:05] [SPEAKER_01]: it's connected up to the other parts of the organization so it's not siloed.

[00:25:09] [SPEAKER_01]: Two, it's interactive and business users can actually play with it,

[00:25:14] [SPEAKER_01]: understand it, use it to make decisions.

[00:25:16] [SPEAKER_01]: And three, properly governed.

[00:25:19] [SPEAKER_01]: Lots.

[00:25:20] [SPEAKER_01]: Basically, everything, everything works out fine.

[00:25:23] [SPEAKER_01]: If you don't do those things, it's exactly like kind of hiring those people into your

[00:25:28] [SPEAKER_01]: organization, it will be a fairly catastrophic failure fairly soon.

[00:25:31] [SPEAKER_00]: And I think that's such an important message to get out there because I do see

[00:25:36] [SPEAKER_00]: so many businesses rushing forward desperate to be a part of that narrative.

[00:25:40] [SPEAKER_00]: Being told just get it over the line rather than risk supporting behind.

[00:25:44] [SPEAKER_00]: And with all that in mind and everything that's going on right now,

[00:25:46] [SPEAKER_00]: how can businesses balance that drive for AI innovation with equally the need to

[00:25:52] [SPEAKER_00]: address potential risks and unintended consequences of AI deployment?

[00:25:57] [SPEAKER_00]: I suspect you've seen a few mistakes made in your time from clients.

[00:26:01] [SPEAKER_00]: I don't expect you to name any names, but what are they doing wrong?

[00:26:05] [SPEAKER_00]: And how can they stop doing it wrong?

[00:26:09] [SPEAKER_01]: Well, perhaps counterintuitively,

[00:26:12] [SPEAKER_01]: our for most businesses, our advice is don't have an AI strategy.

[00:26:19] [SPEAKER_01]: If you're so there are some businesses where their absolute core

[00:26:25] [SPEAKER_01]: competitive advantage is being fundamentally changed by AI.

[00:26:31] [SPEAKER_01]: So take something like outsourced customer service

[00:26:36] [SPEAKER_01]: call center management.

[00:26:39] [SPEAKER_01]: There you probably do need where your AI and business strategy are effectively

[00:26:46] [SPEAKER_01]: identical because you're tackling something so close to the core of what you're doing.

[00:26:51] [SPEAKER_01]: Outside of things like that, let's say if you're a pharma company

[00:26:57] [SPEAKER_01]: that wants to use AI to solve some of your critical problems, maybe something

[00:27:02] [SPEAKER_01]: like improving the efficiency of your clinical trial.

[00:27:07] [SPEAKER_01]: And then you have to think about the other

[00:27:07] [SPEAKER_01]: examples, actually, you should think of this as

[00:27:11] [SPEAKER_01]: you have a business strategy and your AI strategy should serve that.

[00:27:20] [SPEAKER_01]: If you spend too long going around collecting up 200 use possible use cases

[00:27:28] [SPEAKER_01]: of AI in your organization and then prioritize a laundry list of AI

[00:27:36] [SPEAKER_01]: applications independent of your business strategy, what you'll find is

[00:27:41] [SPEAKER_01]: you'll lose energy, your business will lose focus.

[00:27:46] [SPEAKER_01]: And what you eventually build will not affect the core KPIs that you actually care about.

[00:27:53] [SPEAKER_01]: A much better way of tackling that problem is to think hard about your business

[00:27:59] [SPEAKER_01]: strategy and then work with an expert, whether it's internally or externally in AI

[00:28:07] [SPEAKER_01]: and figure out how AI can play a part of that strategy.

[00:28:12] [SPEAKER_01]: Because you actually understand what your business needs to do better than anyone.

[00:28:17] [SPEAKER_01]: If you can find the right expert that can sort of help you

[00:28:21] [SPEAKER_01]: understand how AI can feed into that, well, then you'll be working on the things

[00:28:26] [SPEAKER_01]: that will actually matter to your business.

[00:28:28] [SPEAKER_01]: And so you will have the political energy, the available capital,

[00:28:33] [SPEAKER_01]: the sort of internal resources that actually execute on it.

[00:28:38] [SPEAKER_01]: If you don't do that, you just tend to end up with these POCs lying on the shelf.

[00:28:42] [SPEAKER_00]: And I'm curious in your experience from everything that you've seen now,

[00:28:46] [SPEAKER_00]: any particular qualities that distinguished leaders who successfully navigate

[00:28:50] [SPEAKER_00]: the integration of AI and companies from those who struggle with the transition.

[00:28:55] [SPEAKER_00]: Do you have you observed anything here?

[00:28:58] [SPEAKER_01]: Yeah, it's a very weird mix of sort of arrogance and humility

[00:29:07] [SPEAKER_01]: that seems to me like creates the most successful transformations.

[00:29:12] [SPEAKER_01]: And so it's the humility to really deeply engage with experts that know things that you don't.

[00:29:27] [SPEAKER_01]: Both on the AI side and the people on the front lines of the business,

[00:29:33] [SPEAKER_01]: whatever they like, they will know a great deal of the day to day reality of your business.

[00:29:39] [SPEAKER_01]: And if you can truly engage with those two sets of people while holding the business

[00:29:45] [SPEAKER_01]: strategy clearly in your head, you can develop some really powerful and innovative

[00:29:51] [SPEAKER_01]: solutions that solve the real problems that your frontline workers care about.

[00:29:56] [SPEAKER_01]: But then you need to kind of switch into this slightly more arrogant,

[00:30:02] [SPEAKER_01]: focused mode where there will always be problems.

[00:30:09] [SPEAKER_01]: There will always be blockers inside a real organization.

[00:30:13] [SPEAKER_01]: There will always be vetoes inside a real organization.

[00:30:17] [SPEAKER_01]: And you have to be able to push through those and make sure that the requisite

[00:30:24] [SPEAKER_01]: actions are taken even in slightly difficult circumstances.

[00:30:31] [SPEAKER_01]: And having the judgment to know when to be in one mode and when to be in the other

[00:30:38] [SPEAKER_01]: is also really hard.

[00:30:41] [SPEAKER_00]: Well, it's been such a fantastic conversation with you today.

[00:30:44] [SPEAKER_00]: I've learned so much.

[00:30:46] [SPEAKER_00]: I hope everyone listening will learn so much.

[00:30:48] [SPEAKER_00]: But I'm very conscious we've been looking forward talking about technology,

[00:30:51] [SPEAKER_00]: talking about how it will impact businesses, the problems we're solving.

[00:30:54] [SPEAKER_00]: But before I let you go, I'd love to find out a little more about you because

[00:30:58] [SPEAKER_00]: I think none of us are able to achieve any degree of success without a little help

[00:31:02] [SPEAKER_00]: along the way and very often it's someone that we're grateful towards.

[00:31:06] [SPEAKER_00]: Maybe saw something in us or maybe just inspired us and became our hero.

[00:31:11] [SPEAKER_00]: But is there anybody that you would like to give a shout out today that's had

[00:31:16] [SPEAKER_00]: a significant impact on your career?

[00:31:18] [SPEAKER_00]: Because as I said, none of us get here or where we meant to be without a

[00:31:22] [SPEAKER_00]: little help along the way.

[00:31:23] [SPEAKER_00]: Is there anybody that you'd like to shout out about there?

[00:31:26] [SPEAKER_01]: Oh, I mean, there's been so many people who've been so

[00:31:31] [SPEAKER_01]: transformationally helpful to me.

[00:31:34] [SPEAKER_01]: I think probably, you know,

[00:31:37] [SPEAKER_01]: there's a bunch of people that I'm personally very grateful for my PhD supervisor,

[00:31:41] [SPEAKER_01]: Gabe Epley, my postdoc supervisor at me at Ucobi.

[00:31:47] [SPEAKER_01]: And then some of our investors, Saul Klein and Mark Beath,

[00:31:51] [SPEAKER_01]: but perhaps the one person that was a really significant influence on me

[00:31:57] [SPEAKER_01]: was Charlie Munger, Warren Buffett's business partner.

[00:32:02] [SPEAKER_01]: Like when I was quite a lot younger,

[00:32:05] [SPEAKER_01]: I read poor Charlie Zalmanac and every one of his speeches that I could get

[00:32:11] [SPEAKER_01]: my hands on and throughout the course of faculty sort of steering to what we

[00:32:20] [SPEAKER_01]: thought Charlie Munger would think, as always, always being a really helpful guide for us.

[00:32:27] [SPEAKER_00]: Oh, well, I'll give a shout out to everybody there.

[00:32:31] [SPEAKER_00]: It's so important to get these stories out there.

[00:32:34] [SPEAKER_00]: And for anyone listening just want to find out more information about yourself,

[00:32:38] [SPEAKER_00]: your work, faculty, anything in between.

[00:32:41] [SPEAKER_00]: Where do you like to point everyone listening?

[00:32:44] [SPEAKER_01]: So if it's about faculty,

[00:32:47] [SPEAKER_01]: there's our website, faculty.ai.

[00:32:52] [SPEAKER_01]: I'm on Twitter at Mark 1 of 10,

[00:32:55] [SPEAKER_01]: although to be honest, I probably use it less than I perhaps should.

[00:33:00] [SPEAKER_01]: I tell you what, I'm starting to think about writing up

[00:33:04] [SPEAKER_01]: a bunch of the lessons that we've learned.

[00:33:07] [SPEAKER_01]: So now I've been doing this AI for 10 years.

[00:33:11] [SPEAKER_01]: And so if people would like to

[00:33:15] [SPEAKER_01]: get a hold of whatever form we decide to write that up,

[00:33:21] [SPEAKER_01]: if they drop us an email at info.

[00:33:25] [SPEAKER_01]: faculty.ai will make sure we're on the list whenever we eventually

[00:33:30] [SPEAKER_01]: later in the year bring that out.

[00:33:32] [SPEAKER_00]: That's a fantastic idea.

[00:33:34] [SPEAKER_00]: We will be your accountability partners.

[00:33:36] [SPEAKER_00]: We'll make sure you get that done.

[00:33:39] [SPEAKER_00]: But as I said, love chatting with you today.

[00:33:42] [SPEAKER_00]: You founded faculty to help organizations make better decision using human led AI.

[00:33:48] [SPEAKER_00]: That's one of the things I love about you.

[00:33:49] [SPEAKER_00]: And you were one of the very few London business leaders.

[00:33:52] [SPEAKER_00]: You were chosen to attend that AI safety summit at Bletchley Park.

[00:33:56] [SPEAKER_00]: And in today's interview, just love hearing more about how this grounded

[00:34:00] [SPEAKER_00]: and pragmatic case that you're raising for sensible and so-called

[00:34:04] [SPEAKER_00]: mundane AI regulation and showing your belief that this era of AI has begun.

[00:34:10] [SPEAKER_00]: And over the next decade, every business will not need to become just a tech

[00:34:13] [SPEAKER_00]: business, but an AI business.

[00:34:16] [SPEAKER_00]: So much food for thought.

[00:34:17] [SPEAKER_00]: I encourage everyone listening to send that email and stay up to speed with your

[00:34:21] [SPEAKER_00]: work. But thank you so much.

[00:34:23] [SPEAKER_00]: Thank you very much.

[00:34:24] [SPEAKER_00]: So as we stand on the cusp of this new technological era,

[00:34:29] [SPEAKER_00]: the insight shared by Dr.

[00:34:30] [SPEAKER_00]: Mark Warner today offered an almost powerful roadmap for businesses

[00:34:34] [SPEAKER_00]: navigating the complexities of AI.

[00:34:37] [SPEAKER_00]: And from his belief that every company

[00:34:39] [SPEAKER_00]: must evolve into an AI business to his core for a sensible,

[00:34:44] [SPEAKER_00]: domain specific, AI regulation, I think Mark has perfectly outlined a future

[00:34:49] [SPEAKER_00]: where those that embrace AI safety and innovation, they're the ones that are

[00:34:54] [SPEAKER_00]: going to lead the way.

[00:34:55] [SPEAKER_00]: And it was also great to hear real world stories of faculties work

[00:34:59] [SPEAKER_00]: from life-saving pandemic forecasting with the NHS to enabling ethical

[00:35:04] [SPEAKER_00]: AI deployment. And for me, all of this together demonstrates

[00:35:08] [SPEAKER_00]: the profound impact that well integrated AI can have.

[00:35:13] [SPEAKER_00]: But as you listening, wherever you are in the world,

[00:35:15] [SPEAKER_00]: as you reflect on today's discussion too, I want you to consider

[00:35:19] [SPEAKER_00]: how your business can not only adopt AI, but do so in a way that aligns

[00:35:23] [SPEAKER_00]: with your core strategies and values.

[00:35:25] [SPEAKER_00]: Because the future of business is intertwined with the future of AI.

[00:35:31] [SPEAKER_00]: And the big question is, will your company rise to that challenge or be left behind?

[00:35:36] [SPEAKER_00]: As always, let me know your thoughts.

[00:35:38] [SPEAKER_00]: TechBlogRite.outlook.com, connect with me on LinkedIn at Neil C Hughes.

[00:35:43] [SPEAKER_00]: But until next time, stay curious,

[00:35:47] [SPEAKER_00]: stay informed, keep exploring the ever evolving landscape of technology.

[00:35:51] [SPEAKER_00]: We'll do it all again tomorrow.

[00:35:53] [SPEAKER_00]: But thank you for listening as always.

[00:35:55] [SPEAKER_00]: And don't forget, I'll be back same time, same place tomorrow morning

[00:35:58] [SPEAKER_00]: with another guest. Speak to you all then.