- il y a 2 jours
Business Transformation in the Age of AI
Catégorie
🤖
TechnologieTranscription
00:00Merci.
00:01Merci.
00:10Just a moment.
00:23Bienvenue, tout le monde, ce qui est fantastique.
00:25Nous allons parler de la transformation de l'AIMI.
00:28But before we get into it, just a pause.
00:31I was on the stage talking about AI last year,
00:34and the conversation has shifted.
00:37It is no longer the purview of nerds.
00:41Everyone wants to talk about this,
00:42and every day we see headlines that are one day positive,
00:46this is the future, the next day this is going to destroy the human race.
00:50Before we get to the challenges and the risks,
00:54let's just get from each of your perspectives
00:57where you see the transformation here and the potential.
01:00Let's start with you, Nigel.
01:02Anna, as you said, AI has moved pretty significantly.
01:05We had deep learning and neural nets, and now we're on to generative AI.
01:08The thing that is, I think, a bit of a game changer with this
01:11is that everyday people can now interact with technology
01:15and natural language without them knowing
01:17that they're actually building models
01:19that are helping things get automated
01:21in a way that we've just never seen before.
01:23So while there's a lot of talk about the end of the world,
01:26I feel like we have to deal with things like biases and ethics
01:30and understanding how we keep companies' data and IP protected
01:35and people's data and IP protected.
01:37And the opportunity for transformation in the context of organizations
01:40is with practical things that can now get accelerated,
01:42like risk and fraud in banking
01:45or trying to identify molecules in pharmaceuticals
01:49to advance clinical trials that have been held back
01:52because lots and lots of processing couldn't be done fast enough.
01:55So for me, the opportunity I feel like in the here and now
01:58has just gotten significantly more accelerated
02:01in a way I think that will really move the conversation of AI on.
02:05And I don't think we're done with the development of the technology either.
02:07I think this is the beginning of that way.
02:10Think first websites from the internet back in 95.
02:14Oh, fortunately I can really barely remember that I was just so young.
02:18Nigel, we're going to move on to eventually
02:20how you advise businesses when it comes to this.
02:23But Peter, for Siemens, what is the transformation here?
02:27It's quite niche, I feel like, for industry.
02:29Well, it has specific requirements
02:32because if you want to run a train, if you want to run on a car,
02:36you better make sure that these things are safe.
02:39So therefore, you have extraordinary requirements
02:41with regards to precision
02:42and that these things work flawlessly.
02:45But I do agree with Nigel that, generally speaking,
02:48AI, which was previously with nerds, right,
02:51more in the hardcore R&D developments,
02:54now has become truly mainstream,
02:56which is the reason why everybody is talking around it.
02:58But let me be clear,
02:58we've been using AI since more than 20 years.
03:01We are building CNNs since a long time
03:05and it is almost based in most of the products
03:08that we're using today.
03:09So it's going to be gradually transitioning
03:11into what you really can do,
03:14but to us, this is definitely no news.
03:16Philip, paint us a picture.
03:18Business transformation.
03:20So, and I can't agree with you more, first of all,
03:23around that AI has come front and center.
03:25I used to joke,
03:26I think I was the most boring guy at a cocktail party,
03:28talking about epistemology,
03:30like what do we actually know?
03:31Like what is knowledge?
03:33And then like, let's talk about algorithms
03:34to try and quantify that.
03:36No one wanted to hear about it.
03:37Now I think everyone wants to.
03:39It's really captured the imagination
03:40because this generation is probably
03:42the most approachable generation of technology
03:44we've ever seen.
03:46You know, in businesses of all kinds,
03:47I think the number one thing
03:48that people are really using this technology for
03:50is to connect to their customers.
03:52When I think about customer service organizations,
03:54Orange is here
03:55and they're talking about their use of it
03:57in the call center.
03:58you really are able to answer a question
04:00in the language that the customer has asked it in.
04:02As opposed to having to type in
04:04and find what the answer is in the language of the company,
04:07you can answer it in the language of the customer.
04:10And so in that call center in customer service,
04:12first and foremost, 100%.
04:14The other area is really in marketing.
04:16Organizations, every organization I talk to says,
04:19if I could just connect to that audience better,
04:22or I could just connect to that demographic better,
04:24or I could just connect with this type of a message better.
04:27And it's hard to repetitively be able
04:29to create that information.
04:31Carrefour is a great example that's here as well
04:33where they're actually creating marketing content
04:35by audience that they want to try to connect to.
04:38And so I see this generation
04:39as putting a conversational interface,
04:42the interface that we were designed to have
04:43and just having a conversation with our information.
04:45That's what I think a lot of organizations see
04:48as the opportunity for this generation.
04:50It is tricky though, I think,
04:51for a number of businesses to know how to adopt AI.
04:56Some businesses are frankly putting their head in the sand
04:59and just banning employees from using it.
05:02Big fear about proprietary information being shared.
05:04But Nigel, how do you advise companies that come to you
05:06and say, we want to get into AI,
05:08we want to do it fast, it's a journey?
05:12For sure.
05:13But you have to start by experimentation.
05:15One of the biggest challenges
05:16with putting something like ChatGPT out into the world
05:19and companies not having a way to test information
05:22in a secure sandbox
05:24is everybody throws your confidential information
05:26to ChatGPT.
05:27The model is getting smarter,
05:29but it's getting smarter off your data.
05:30So the first thing we do with a lot of clients
05:32is establish, you know,
05:34we've got OpenAI and Azure Cloud environments
05:36that we establish proprietary sandboxes
05:39so you can actually start training the model
05:41on your data that stays within your wall.
05:43So we partner with the likes of Car4
05:45on their digital transformation.
05:47And if you're starting with a company like that,
05:48you want to start by establishing sandbox
05:51where people can experiment.
05:52But the other thing that, you know,
05:54we've just talked about, right,
05:55AI is not something that is going to drive business transformation
05:58or technological transformation and isolation.
06:00You still need the other skills like product,
06:03like experience and how you design things
06:05and engineering and being really clear
06:07about what is the use case
06:08from a strategic perspective
06:10that you want to get out.
06:11We use this acronym called SPEED.
06:13Strategy, Product, Experience, Engineering, Data, and AI.
06:17And if you, you know, it's like fingers in a hand.
06:19If you go with a strong thumb
06:21or a strong index finger,
06:22it doesn't matter if it doesn't connect to the pinky.
06:24You still can't move and lift and shift stuff.
06:27And that's where a lot of companies get slowed down
06:29because they say, yeah, we have these,
06:31we have engineering, we have experience, we have AI.
06:34We've been doing it for 20 years,
06:35but they're not connected in a way
06:37that actually is allowing them to move forward.
06:39And today, as Philip just said,
06:41customers are using natural language to talk
06:44in a way that they can push you harder than they've ever had.
06:49You know, so accepting marketing messages that are generic,
06:53they don't want that.
06:54You know, in a quick service restaurant,
06:56getting the order wrong
06:57because you couldn't understand what they were saying.
06:59They don't like that.
07:01And companies who start to figure out these use cases
07:03in the here and now
07:04will make significant progress.
07:06So experimentation is where we advise our clients to start.
07:09And also, I think with general sort of AI platforms,
07:14that doesn't really work for a lot of businesses.
07:16I've been really enjoying some terrible use cases of AI.
07:22For Siemens, it must be quite interesting
07:23because you have such specific use cases for AI.
07:26Is it ever hard to figure out what platform you need to use?
07:31Really, it is.
07:32I mean, first of all, you are right.
07:34It is at the very heart of what we do.
07:36So if you think about, I mentioned trains, right?
07:39We are able to run trams autonomously.
07:42I mean, we're talking about self-driving cars,
07:44but you can do it already on the railroad
07:46because think about it, it's going forward, backward.
07:48So it's much easier in that sense to start there.
07:51We talk about energy transition.
07:52We talk about green energy.
07:54As we move in that green energy space,
07:57you need to have AI that actually really stabilizes the grids,
08:00which we built already into it.
08:02Or think about factories, right?
08:04Where we think about picking and place.
08:08So things in the warehouse.
08:09Today, it's very repetitive.
08:11It's very hard for workers to do.
08:13Now, we can train robots really, really easily,
08:15you know, based on AI,
08:17actually to do it very, very precisely,
08:20and so therefore liberating, of course, the factory workers.
08:24So all of this is happening already as we speak.
08:27Now, the nice thing about generative AI is two things.
08:30Number one, your language.
08:31So for customer service, it's great.
08:33For marketing, it's great.
08:35For actually tearing down cultural differences
08:38when it comes down to languages so that people do understand.
08:41You have a service ticket.
08:42It's automatically then translated,
08:45then sorted, and then actually put back into the database.
08:49Perfect.
08:50So that really works well.
08:51And now what we're experimenting with
08:53is can be used for coding
08:54because we get 24,000 software engineers.
08:58And think about the potential
08:59if you were able now to use it for code generation.
09:03So there we're not quite sure yet.
09:04We look at it as it can do wonderful documentation.
09:08It can do wonderful validation and testing.
09:11But for code generation,
09:13we definitely need to find platforms,
09:15as Nigel was explaining,
09:17where we may have to reinforce it
09:19in terms of the learning
09:20so that we get this right,
09:22which probably will be more proprietary
09:24than actually as a major platform for everyone to use.
09:27Billup, we were talking on the phone last week
09:29about generative AI
09:30and running through some fairly amusing cases
09:32of it going really wrong.
09:34Hallucinations,
09:35love a weird chat with a chatbot,
09:37but inaccurate information.
09:40Do we think this is teething problems?
09:43Is it the problem of the AI?
09:44Is it the problem of us using it wrong?
09:47Yeah.
09:47I think these algorithms
09:49are like confident toddlers.
09:51They'll just blurt out the last thing they heard.
09:54Sometimes it's about your mother-in-law,
09:55and you're like,
09:56where did it hear that from?
09:58And so it literally does assemble an answer.
10:02In medicine,
10:02in some cases,
10:04they're taking two Latin words,
10:05putting it together,
10:06and saying you should treat this,
10:07and the medicine doesn't exist on the planet Earth.
10:09It just makes it up.
10:10But it thinks that's what you wanted.
10:12Well, the algorithm is designed to give an answer.
10:15That's its job,
10:16is to give an answer.
10:17And so humans have to intervene in this.
10:20The training that you have to use
10:21has to intervene.
10:23If you really can constrict
10:24what the inputs are,
10:26you can predict the outputs of this algorithm.
10:29And so there's a whole variety of techniques
10:30that you can use.
10:31We believe,
10:32you know,
10:32a lot of the technology we've been showing
10:34is combining together things like search
10:36with these large language models.
10:38So when it gives an answer,
10:39it says,
10:40this is where I got the information from.
10:42And you do citation.
10:43And it says,
10:43this is how I assembled it.
10:45We're even doing things with images
10:46where we put a watermark inside of the image,
10:49and we have a metadata specification
10:50so I can say,
10:51this was generated by an AI.
10:52So I know if it was generated by a human
10:54or generated by an AI,
10:55and so the content
10:56that you put into these algorithms
10:58is critical.
10:59Additionally,
11:00we're working on a lot of technology
11:01that we're releasing
11:02to actually use the AI
11:03to watch the AI,
11:04to make sure it stays on task.
11:06That's very massive.
11:07And the model hasn't been poisoned
11:08or it hasn't changed,
11:09it doesn't skew.
11:10And so we will use the technology
11:12to make these models better
11:13and better and more accurate,
11:14but it really starts with the foundation
11:16that you give to these models.
11:18Okay,
11:19let's talk about jobs.
11:20This is something you were touching upon,
11:21Peter,
11:22and it's not a particularly popular conversation.
11:26There are huge efficiencies here
11:28for businesses
11:29when it comes to jobs
11:30because AI can perform so many tasks.
11:34Is this something we should see
11:35as a positive or a negative,
11:37Peter?
11:38Yeah,
11:38let me start there.
11:39We clearly see it as a positive
11:40for two reasons.
11:42Number one,
11:43we have a severe shortage in skills.
11:46we know that,
11:47for example,
11:48in China,
11:49the next 20 years,
11:50140 million people
11:51will go out of the workforce.
11:53We have similar situations
11:54in Europe,
11:55in Germany,
11:56in France,
11:56in Italy.
11:57If you think about it,
11:58it really is a major issue.
11:59So we really have to be productive
12:01for the future
12:02as our populations are shrinking.
12:04So labor shortage is a big issue.
12:07Second though,
12:09think about what I was talking about,
12:10the factory workers liberating them
12:12from mundane jobs
12:13where they are overly repetitive
12:15and they're really a big strain
12:16on your psyche as well.
12:18How wonderful isn't that
12:19that if you could use AI
12:20in order to make this work?
12:22However,
12:23in our learnings,
12:24and we've done a lot AI
12:25in medical actually,
12:26and I do agree
12:26that of course,
12:28it needs to be precise
12:29and it needs to be validated
12:31and all of this,
12:33but it needs to be,
12:34the key word is companion
12:37because the companion word is,
12:38I'm not threatening you.
12:40I'm here to support you,
12:41but still,
12:42you as a human,
12:43you are in charge.
12:44you make the final decision.
12:45It's a companion.
12:47And I think a lot will come
12:49with the way we position it,
12:50we really say,
12:52you know,
12:52this is here to help you.
12:54And I don't think
12:54that right now
12:55the conversation is about,
12:56oh, it's going to kill
12:56that many jobs.
12:57It's going to particularly
12:58help for both the conversation.
13:00Do you see the workforce
13:01that's even shrinking
13:01as a result of AI?
13:03I think the workforce
13:05will become more productive
13:06provided that we all,
13:09lifelong learning,
13:10upskilling,
13:11that we actually get adjusted
13:12to it.
13:12But that depends a little bit
13:13on actually the usability
13:15of the technology.
13:16If it's easy to use,
13:17which is the reason
13:18why we talk about
13:18large language models
13:19right now so much,
13:20although AI has been around
13:22for the longest time,
13:23is because it became mainstream
13:24because it's easy to digest,
13:26it's easy to access,
13:27but therefore now
13:28we have to put the guardrails
13:29in place
13:30that people understand
13:30where it's coming from,
13:32what it can do,
13:32but also where
13:33its limitations are.
13:34Nigel, you have the inside track
13:36really because you speak
13:36to so many business leaders
13:38about how to adopt AI
13:40into their strategies.
13:42When you speak to these leaders,
13:43what are they looking for?
13:45Is it faster growth,
13:47efficiency,
13:48is it cost-cutting?
13:49What's top of mind?
13:50I think when you think
13:51about the situation
13:52we have right now,
13:53a high inflationary environment,
13:54supply chain challenges,
13:55most businesses
13:57are looking to drive growth
13:58whilst being more efficient.
14:00It's not an and,
14:01it's an, you know,
14:02you have to do both
14:03at the same time.
14:04And one of the big advantages,
14:06as we're just talking about,
14:08I think with the push around AI
14:10is you can start to bring
14:11more people into the workforce
14:12who are starting to perform
14:15technological tasks
14:16who were not technologists
14:18to begin with.
14:19Think about the power of that,
14:20Think about low-code
14:21and no-code models
14:23being able to bring people
14:24who essentially are smart
14:25but have never done
14:27a computer science degree
14:28can now start to produce outputs
14:30that meaningfully advance you.
14:32Also, I think this notion
14:33of actually a co-pilot, right?
14:35You're not basically saying right now
14:37that this thing's going to fly the ship.
14:39You're saying it's a co-pilot.
14:40It's advancing you.
14:42So it helps business leaders
14:43in particular bring people
14:45into this digital sphere
14:46that was pretty gapped
14:48from a talent perspective.
14:49If you just look at
14:50every Western economy,
14:52there are short skills
14:54primarily in the technology sector,
14:56even when we have
14:58an excess of people
15:00who have not been able
15:02to find roles in other sectors.
15:03So the potential for them
15:05to now enter a space
15:06and start to say,
15:07hey, I can actually,
15:08with limited technology learning,
15:11start to use tools
15:12that allow me to progress
15:14in this new space
15:15from designing
15:16to writing code.
15:17I have a 13-year-old
15:19who's now started
15:20to write code
15:20way better than he ever did
15:22and build real things
15:24on Roblox for gaming
15:26using platforms like GitHub
15:29to basically produce
15:30most of the code.
15:31And I was thinking about this.
15:32You know, it's like,
15:33I kind of,
15:34I'm a little older than you.
15:35I guess I go back
15:36to the early days
15:36of the internet.
15:37And this is a little bit
15:39like what happened
15:39when you had to write
15:40notepad code
15:41to write web pages
15:42in HTML.
15:43And all of a sudden,
15:44people started to give you
15:46WYSIWYG editors
15:47and you were like,
15:47holy smokes,
15:48I can produce this thing
15:49way faster.
15:50So I think he's going to
15:51skip that whole generation
15:52of code writing
15:53that many developers
15:54are being put through today
15:55by simply learning
15:57how to do it
15:57from the get-go
15:58with a co-pilot.
16:00What do you think, Philip?
16:01Is this just going to
16:02transform the next generation
16:04of jobs?
16:06100% it is.
16:07You know,
16:07when we invented
16:08the printing press,
16:09all of a sudden,
16:10the eyeglass industry
16:11was invented
16:12because people didn't realize
16:13they were farsighted
16:14until they had to read.
16:14We invented the air conditioners
16:17to be able to keep
16:17those printing presses cool.
16:19And then we started
16:19using those air conditioners
16:20inside of a place
16:21called movie theaters
16:22and people started
16:23flocking to movie theaters.
16:24And so a simple invention
16:25of actually just making it
16:27easier to produce words
16:28and generated
16:29so many new inventions
16:30and so many things
16:31that we take for granted today.
16:32You know,
16:33and in this era,
16:34you know,
16:34I agree with everything
16:35that's been said,
16:36which is, you know,
16:37I look at the productivity gains
16:38and the world champion
16:40in chess is not an AI.
16:42The world champion
16:43in chess is a human
16:44aided by an AI.
16:45And that's one
16:45of the longest serving problems
16:47that we've worked on
16:48inside of AI.
16:49And, you know,
16:50organizations,
16:51I look at productivity gains
16:52of somewhere around,
16:53somewhere between
16:5410 to 30% on average.
16:56The more expert you are,
16:57the closer it is to 10%.
16:58The less of an expert you are,
17:00it's closer to 30%.
17:01And so that means
17:02that all of us,
17:03when I think about
17:04being able to give
17:04all of ourself
17:05a day back of productivity
17:06a week,
17:07being able to take care
17:08of 20 or 30%,
17:0925% of your emails,
17:10or not having to respond
17:12or write up those
17:12meeting numbers,
17:13the things that are like
17:14very, like make work drudgery,
17:16I think is the opportunity
17:18to really make work
17:19exciting again
17:20and make it fun again.
17:21And so, you know,
17:22the areas that we talk
17:23a lot about is
17:24look at the places
17:25where you have a hard time
17:26retaining people
17:26and hiring people.
17:28That tells you
17:28there's something about
17:29that job
17:29that's not rewarding.
17:30So how do you either
17:31take the drudgery out
17:32and make it rewarding?
17:34Second area is really
17:35in the area of like,
17:36if I've got to create
17:36more content
17:38to be able to get more sales
17:39and I'm gated,
17:41RFPs, RFIs,
17:41emails that have to
17:42go out to clients,
17:43how can I help
17:44accelerate that?
17:45I can generate more business.
17:46And the last is
17:47your highest paid talent,
17:49those individuals
17:49that you consider
17:50your experts.
17:51What do you have them
17:52doing over and over again
17:53that's not adding value
17:54to their job?
17:55When you look at those areas,
17:56I think everyone says,
17:57wow, if you could take care
17:58of just 20% of my work,
18:0020%, 25% of my work,
18:01I would do a copilot
18:03or a duet
18:03or use AI
18:05any day of the week.
18:07So far on this stage,
18:08we've just heard
18:10so much positivity,
18:11so much potential.
18:13But there are risks
18:15to AI.
18:15And there's a huge debate
18:17about how much
18:18this technology
18:19should be regulated.
18:20Just yesterday,
18:21the EU Parliament
18:22approved the draft text
18:23for an AI Act.
18:24What do we feel
18:26about the almost endless
18:28barrage of warnings
18:29from some of the biggest
18:31minds of AI,
18:32the pioneers,
18:34people like Elon Musk
18:36calling for a pause.
18:38Are the fears overblown?
18:39Nigel?
18:40You know,
18:41from my perspective,
18:41I think the fears
18:42of the end of the world
18:43are much further away.
18:45And as you were pointing out,
18:46the fears of here and now
18:47are more important.
18:49How do we make sure
18:49that the world of AI
18:50doesn't become as exclusive
18:52as the early digital world?
18:54How do we make it
18:54more inclusive?
18:55How do we think about
18:56ethics and guidelines
18:57for how AI should be set up?
18:59How do we think about
19:00bias and hallucinations?
19:02Because, you know,
19:02these are problems
19:03of the here and now.
19:04And for me,
19:05us addressing those problems
19:07creates the platform
19:08on which you can start
19:09to regulate
19:10the bigger conversations.
19:12My worry is
19:12we focus so much
19:13on the end of the world,
19:14which may or may not come
19:15in a 10-year period,
19:17and we ignore the problems
19:18of today,
19:18which are going to affect
19:20millions of people right now.
19:21You know, deep fakes.
19:22We have elections coming up
19:23in about seven countries
19:24over the course
19:25of the next year.
19:26If you don't give people
19:27a way to start to parse
19:29through understanding
19:30what's the origins
19:31of the content,
19:32is it genuine,
19:33is it not?
19:33It could have massive
19:35consequences for politics
19:36and government
19:37and choice now.
19:39And those governments,
19:40if they're, you know,
19:42there because they've
19:43used technology and AI
19:44to their advantage
19:45to move people
19:46into thinking things
19:47that were not true
19:48were true,
19:49will be doing us
19:50a disservice
19:50over the 10 years
19:51that we have to figure out
19:52how to deal
19:52with the end-of-the-world problems.
19:54So for me,
19:54I think there's a timing issue
19:56that's extraordinary,
19:57which is focus
19:57on the pressing issues
19:58of now,
19:59and then let's have
20:00a more mature conversation
20:02as people understand
20:03the technology
20:04over a period of time.
20:06Peter,
20:06is the answer
20:07policy and regulation?
20:09I mean,
20:09we're both Europeans.
20:10The EU is a leader
20:11when it comes
20:12to tech regulation,
20:13possibly not tech.
20:14Is this the answer?
20:16Well, right now,
20:17definitely we see
20:19a lot of papers
20:21being published
20:21and then,
20:22as you said just yesterday,
20:24the EU AI Act
20:25coming through
20:26after GDPR,
20:27after, of course,
20:28the EU Data Act,
20:29everything that we're discussing.
20:31We need to be careful.
20:32In particular,
20:34given the pace of technology,
20:36because think about
20:37how long it takes legislature
20:38to actually come up
20:39with a law
20:40that then already
20:42is going to be outdated
20:42by the time the technology
20:43really has advanced.
20:45So we need to make sure,
20:46of course,
20:46that the consumer
20:47is safe.
20:49Safety is absolutely
20:50non-discussable.
20:51So that's fine.
20:52But then we still need
20:54to hold people accountable
20:55and companies accountable
20:57that if they use AI,
20:59you are on the hook.
21:00If we at Siemens,
21:01we provide you services
21:02that are AI-based
21:03and are not working,
21:04guess what?
21:05It is Siemens' accountability
21:07and liability
21:07to make sure
21:08that this is the case.
21:09So we need to be
21:10a little bit careful
21:11because if you,
21:11we are competing
21:12on a global stage
21:13and we are discussing
21:14about how we Europeans
21:15actually will see
21:17more tech in Europe.
21:19And I don't believe
21:20that regulation
21:20will be per se the answer.
21:22It is great
21:23that we are leading that,
21:24but how about
21:25we really create
21:26these jobs,
21:27these products,
21:28platforms
21:29that we can pull
21:30across the globe
21:31because we are in competition
21:32with China,
21:33we are in competition
21:33with the United States
21:34and that I think
21:36we have to be
21:37really careful
21:37with that.
21:38You get the last word.
21:41So is the world
21:42going to end?
21:43Should we pause AI
21:44and is the answer regulation?
21:46You've got two and a half minutes.
21:47Two and a half minutes.
21:48Well, first of all,
21:49I want to compare this
21:50to blockchain and Bitcoin.
21:52It was invented back in 2010
21:53and think about
21:54we're finally arriving
21:56at some of the regulations
21:57related to that technology.
21:59We regulated
22:00and we managed
22:01that technology,
22:02I would tell you,
22:03poorly for way too long.
22:05We are thrilled at Google.
22:07We invented
22:08the transformer algorithm
22:09back in 2017
22:10which underlies
22:11a lot of this core technology
22:12and in 2018
22:13we built a set of principles
22:15inside of Google.
22:16Things like being
22:16beholden to humans,
22:18not reinforcing unfair bias,
22:20privacy by design
22:22and we knew then
22:23what the opportunity
22:24and also what the risks
22:25were of this
22:26and so we implemented principles.
22:27We're very much
22:29in favor of regulation
22:30like I would say
22:31smart regulation
22:32but regulation
22:33that doesn't stifle innovation
22:35agreed 100%
22:36but we're really thrilled
22:37that the whole world
22:38this early on
22:39in developing this technology
22:40I would tell you
22:41it's only been released
22:42to the world
22:42in about the past year
22:43and already the whole world
22:45is talking about
22:45using this responsibly
22:47and so no, I don't.
22:48I have more faith
22:49that if you can write
22:50a line of code
22:50that does something bad
22:51somebody can write
22:52a line of code
22:52to actually protect you
22:53against it
22:54and I'm glad
22:54that we're having
22:55the conversation
22:56so the technologists
22:56actually have guidance
22:57on how to do it.
22:58and I think also
22:59what's so positive
23:00is that the pioneers
23:01of this technology
23:02are actually leading
23:04the conversation
23:04and are speaking
23:06to governments
23:06and there is a conversation
23:08being had
23:09so hopefully AI
23:10can be all about
23:11the potential
23:11in the future
23:12and less about the risks.
23:15Perhaps when we're back
23:15here next year
23:16the conversation
23:18will have moved on
23:18I'm sure once again.
23:20Thank you very much
23:20to all three of you
23:21that was wonderful.
23:22Thank you.
23:22Thank you very much.
23:24Thank you very much.
23:25Thank you.
Commentaires