Rise of AI Conference 2024 · Berlin · Applied AI Stage

Ready for the Era of AI?

Irina Gontcharova · Director of Azure Infrastructure, Microsoft Deutschland GmbH

At Rise of AI 2024 in Berlin, Irina Gontcharova, Director of Azure Infrastructure at Microsoft Deutschland, argues that AI does not solve human problems by itself but unleashes human creativity, much like the invention of the collapsible paint tube in 1841 freed painters to leave the studio and enabled impressionism. She traces how fast AI adoption is moving: ChatGPT reached mass adoption in about two months, far faster than mobile phones or the internet. She lays out what she calls the AI stack organizations need to get ready: infrastructure such as GPUs, a working data platform, application tooling built on top of models, and security and governance running through all of it. She stresses that Microsoft builds Copilot, not autopilot, keeping a human in the loop and in control. Using a live demo with an AI-generated balloon image, she shows how a model reasons about physics questions and where it still hedges or hallucinates. She closes by saying that readiness for AI requires imagination, which cannot come from a machine, combined with a responsible approach.

Key takeaways

  • Gontcharova compares AI to the 1841 invention of the collapsible paint tube: a tool that unleashes human creativity rather than solving problems by itself.
  • ChatGPT reached mass adoption in about two months, far faster than the internet or the mobile phone.
  • She structures AI readiness as a stack: infrastructure, data platform, application tooling, and security, with policy and education running alongside it.
  • Microsoft builds Copilot rather than autopilot so a human stays in control of AI-assisted work such as coding.
  • A live demo shows a model reasoning about a balloon image with hedged, probabilistic answers, and that the same question against different data sources produces different results.

Detailed notes

A paint tube for the age of AI

Gontcharova opens with an exercise: naming what was significant about the year 1841. Search engines and copilots surface a US presidential inauguration or an Antarctic expedition, but she points instead to the invention of the collapsible metal paint tube. Before it, painters carried pigment in pig bladders that could not be resealed, so they worked in the studio from memory and sketches. The tube let them paint outdoors directly from life, and she credits it with unleashing the creativity behind impressionism and painters such as Monet, Renoir and Pissarro. Her point: AI, like the paint tube, does not solve human problems on its own. It is a tool that unleashes human creativity, and the imagination and judgment still belong to the person holding it.

Adoption is moving faster than any prior technology

She contrasts adoption speeds: it took the mobile phone about 16 years to reach 100 million users, the internet took less time, and ChatGPT reached comparable adoption in about two months. She frames this as a democratization of technology, comparable to how smartphone cameras let anyone create content without being a trained painter. She cites Microsoft investing 2.3 billion euros in AI in Germany and a large skilling initiative aimed at people in the country, though she is visibly unsure of the exact figures while speaking. She also cites an internal Microsoft study finding that AI technology investment pays back within roughly 14 months, which she frames as a deliberate choice companies in tight financial situations have to make.

The AI stack organizations need to get ready

Gontcharova breaks AI readiness into layers: infrastructure (GPUs and InfiniBand networking, from vendors such as AMD or Nvidia); a data platform, where she says few organizations manage to connect the people who build applications with the people who hold the data; AI platforms and toolkits for building and extending applications without being limited by scarce development resources; and security, which she says is non-negotiable because the same generative tools that create value can be used by bad actors to run scams. She notes GitHub Copilot can write code but is designed to work with a developer, not replace one, and that Microsoft deliberately calls its products copilots rather than autopilots for that reason. She adds that government policy and guardrails, plus education from K-12 through universities, have to be part of the same conversation, since most people preparing for this shift will not have studied computer science.

Demonstrating reasoning with a balloon image

She walks through a live example of asking Copilot to reason about a DALL-E generated image of balloons on strings: what would happen if the strings were cut, what gas is inside the balloons, and whether hydrogen would work at a barbecue. The model correctly infers the balloons would float away due to buoyancy, judges the gas is unlikely to be air and more likely helium or hydrogen, and warns against using hydrogen near a barbecue because of its flammability. She uses this to show the model reasoning with uncertainty ('unlikely', 'I'm not sure but') rather than stating facts outright, while noting that hallucination remains a real risk in these systems.

Different data produces a different answer

She recounts asking Copilot to analyze her own communications and say which superhero she resembles, getting 'Iron Man'. When a colleague asked the same question of their own data, they got the same answer, which she found unconvincing. Feeding the model a different dataset, a fandom wiki, changed her result to 'the Oracle from The Matrix'. Her conclusion: the answer a model gives depends entirely on the data connected to it, so organizations have to work out deliberately what data their AI applications should draw on rather than expecting a standard model alone to produce useful results.

Closing message

Gontcharova closes by saying no single company or country can do this alone and that partnership is required across the ecosystem. She summarizes what readiness for AI requires: imagination, which cannot come from a machine, and a responsible approach, framing these as the two things that let people 'paint the future' the way the paint tube let Monet, Renoir and Pissarro paint theirs.

Quotes

It's not the AI that solves the problem, it's us, human, holding this paint tube.

Irina Gontcharova

You need imagination, a lot of it, because that's nothing that can come from machine, and you need a responsible approach.

Irina Gontcharova

Questions answered in this talk

How fast did ChatGPT reach mass adoption compared to earlier technologies?

According to Irina Gontcharova at Rise of AI 2024, ChatGPT reached comparable adoption levels in about two months, compared with roughly 16 years for the mobile phone to reach 100 million users.

What layers does an organization need to get ready for AI, according to Microsoft?

Irina Gontcharova outlines infrastructure such as GPUs and networking, a working data platform, AI application tooling, and security and governance, on top of which organizations need imagination and a responsible approach.

Why does Microsoft call its AI products 'copilot' instead of 'autopilot'?

Irina Gontcharova explains that Microsoft deliberately keeps a human in control of tools like GitHub Copilot, using them to assist a developer rather than replace one.

Mentioned

Generative AI, AI infrastructure, Data platform, Copilot vs. autopilot, AI adoption speed, Responsible AI, Critical thinking

Keep exploring

More from Rise of AI Conference 2024

More on Enterprise AI & Adoption

Transcript

Auto-generated transcript, not edited. · 4,373 words

Show the full transcript

thank you so much I am absolutely excited to be here today and so let's check out if I'm actually capable of um you know is using some simpler technology here so welcome Berlin um fantastic two speakers is really I was feeling like in my brain I was going like plus one plus one plus one we need more skill uh people we need more people who understand the ey we need to talk about things controversially because that's the only way how we humans learn from each other and that's how we can make whatever technology has at our hands better for the future um and I'm pretty sure you want to have some future predictions I am about to disappoint you terribly because I'm actually going bad so for those of you have a co- pilot at your hand you can try this exercise with me I'm going going back to the year 1841 so 1841 and if you want to do this mentally you can think for yourself of any historic event of any significance um

to that particular year so I did run that a couple of times that's not the first time I presenting that and if you ask um co-pilot or you go into web search whatever choose uh you will come up with some things uh you will anybody from us here I think it was a 10 in regation for the president um John Tyler I have to look up that I never remember and there was an Amistad case actually trit in US court of freeing people who were kidnapped on the board of amistat um there was Antarctic expedition figuring out some some new Geographic uh capabilities there but this is not what I think been actually most important that goes back to Katrina's point that my personal judgment neither of these web responses are or co-pilot responses is actually touching my nerve what does touch my nerve is this invention here and um you might guess what that is and because we have limited time I don't give you too much time for the guessing this is actually

a paint tube it's a sink paint tube that you can screw in now here is why I believe this was a one of very important inventions in a human history before that paint tube has been invented and You Were A Painter in 184 you w or before that and you would pick a pig blooder and you would put your paint in there if you wanted to travel and if you couldn't do that because Pig BLS trust me on this one I haven't tried but I believe that are very terrible in transporting paint and once you open it you couldn't close it so it was a very strong limitation to the creativity of any painters so what these guys ended up is they were painting in studio so they would go out they will capture whatever they could on a sketch that will memorize what they've seen in terms of the lives or situations or Impressions on faces and then it will come back and then it will paint so the invention and the squeeze of

that tube imagine what that done it actually Unleashed an evolution of creativity and now we be great of what we call today an impressionism there will be no rinoa there will be no song there will be no goong any of these great paintings if there would be not this little simple pain tube why are we talking about this today because I do believe that often times and tan canina touch on that multiple times in day talk track we actually refer to AI such as this would solve all the human problems at once and it wouldn't but it does unleash a human creativity to the immense really that it was done here because all of the sudden painting a picture of a mother or people in a park at a picnic was so much more real and believe it or not we will see many many of those Innovation coming up why I'm telling this so this was a patent find by John John Gand in 1481 it was one patent that Unleashed this Revolution and

painting let me take you forward there are 22,000 pain like patents by Microsoft only and many more and Tina had some stats on that as well and I'm pretty sure we can find many more in line there are millions of patents actually 22,000 already on Microsoft 10,000 in Germany if I remember the numbers correctly from the previous speaker eight Global Research centers here 9 in Germany thousands and thousands of people creating what we imagine a pain tube today on AI space why it is important because it's not the AI that solves the problem it's us human holding this pain tube and I want us to remember what it does it enables the evolution not of a human race or artificial intelligence and its whole it gives us the means to creativity to get there faster to be more real and to be outside of the studio versus in studio so I started my career as a computer scientist so I can very much connect to what Katrina was saying before that and I remember my

first Professor telling me Arena machine only does what you tell it to do well you know fast forward certain years later it's still true the only difference today is that machine is much more powerful so I can interact with a native language I have a gup copilot at my fingertips so I can be faster in creating whatever imagination I think but all of the principles and rules apply it's still our imagination it is still our creativity but what it does and I think that's that's the fundamental difference between the technological innovation we've seen before and technological innovation we've seen since um a couple of I like months now really the pace of adoption has changed fundamentally it took us whatever number of years for the mobile phone was 16 years to get from one user to 100 million it took us an internet s years so for those folks who was in computer science as I assume most of you were we were pretty quick there because we've seen the potential but you know like

my mother didn't have an internet for many many many many years now she does um it took us two months at the moment when chbt came out to for adoption what it does it democratizing a lot of that technology so everybody has you know not everybody was a painter back there in 1941 everybody can take a picture of of a phone or can go and create content online because of the iPhone momentum today with generative a and many tools that come after that which we're going to talk about in a second that's what we kind of getting we are getting to the many many many people can have access to technology and that's why we have the responsibility also to provide the literacy to that technology as Kine R said now Microsoft t St on where we are at is we truly believe and I'm here for that very same reason not just here today but with that company for many many years is we want that every individual every team every industry does go

and does discover the power of that creativity the power of what lies behind underneath provided by research and people who get that into action we don't need everybody to be skilled scientist to reason with the model but we need to provide the people basic literacy not in computer science but frankly in critical thinking in smart questioning there was a horard business review um article very recently and if if you're interested catch me on a cafe break I will figure it out for you um that said that accumulation of knowledge and synthesizing the knowledge still super important for humans but the real true Advantage comes when you start asking critical questions so if you ask me and I'm a mother of two what is the most important skill I want my kids to be able to have is the critical thinking and smart questioning and being able to reason both with other humans equally with a computer generated whatever ever that is now when we take this forward we actually really need to get that down

to Action Now how do you get that dark to action because this AI transformation doesn't happens all by itself it happens because all of you here all of the people in the other stage all of the people out there everybody in Germany or wherever else in the world it starts to think of practical applications enriching employee experience like giving me I'm a knowledge I consider myself a knowledge worker now yesterday Gary asked me this he said like aren are you still in knowledge work I'm like yes absolutely because my power is this brain and and chity is has access to way more much information that I do but I have two degrees in critical thinking that H doesn't so what I can do is I can imagine in my daily work and I work with a calet every day figuring out how I can make my work life easier with is the assistant that is available to me today at the fingertips so I'm already using a lot of this assistanc thankfully now the customer

engagement is now different things now whether I am able or not able to set a dynamic pricing but what I expecting from this future transformation that that whenever I interact with a with a an e store or a customer service or I'm calling my telephone provider that the people on the other line do not jump me three times of of Hoops through the same questions because it just gets frustrated right so that's a capability of generative technology that is there and can be adopted reshaping business and more interestingly moving these frontiers of innovation because we don't hold ourselves back by and dragging our feet by executing basic regiment tasks now this is what we're trying to do together all of you and us as in Microsoft and by the way I I hear all of this like big us um there are millions of people working for Microsoft all over the globe uh just because the company headquarters in rment I consider myself my Microsoft Germany we just invented 2.3 billion into this particular industry

Skilling 1.2 I don't like remember 12 million people here in Germany in order to be able to be moving this particular society forward the reason why I'm so proud of what we're trying to is the why and let me take you I hope that works now let me take you this this does it it doesn't no it doesn't okay so I'm going to share with you the video so let me so the video is actually talking about the providing an opportunity for the people to trust themselves for what they dreams are with a Cod it it is now here is what I think is super important we talk and that goes back to the point what is AI and how do we get ready there because that's what I promised you to tell and to talk about so please go ahead and ask these questions AI is not just one model it's a fundamental stack of pieces of information and a pillars that we should go and talk together whether you're a startup whether you're

developer whether you're work for government whether you're an industry representative these are the things that we need to talk and exercise ction smartness and execution at any of these levels now infrastructure for AI is a pretty straightforward topic infrastructure is a GPU infin band whether it's AMD or Nvidia giving a fun fact the first jgpt model that came out 2022 um the capacity that you needed in order to train that was 285,000 AMD Infinity band computers and that was like 10,000 Med v00 I believe uh to train that one model which is we all agree is super fency and very very fun but not smart enough to do multiplications as we just learned um but there are more things that needed in order for eii to be meaningful to us and to really unleash the potential that we all have in US humans hopefully the positive one not the negative one and it's a data platform and I I have an example for you just in a second about the data platform now we talk

to a lot organizations I am surprised by how few actually managed to cross over the silus to make sure that the people who need application talk to the people who have the data and vice versa I also had this one interesting conversation where people say like yeah if I connect everything in one Lake and then then I unleash the CH on this then I unbias my data I'm like that's an interesting conversation I don't think so because I am of the opinion and we can have the research conversation about that how much bias is or in that data right so we have to be smart of first actually enabling chpg to to access that I tell you why in a second and then beyond that this is all not for self- reasoning right this is all because the real application starts when you have an application that helps a particular business problem in a particular business to be solved whether you're in a medical care and you need to have a roster in schedule or

or for a patients or you have um I don't know do bonds somebody here um to to um optimize your train schedule or your cruise for the flights or whatever it is there are multitude of these problems but they all go with applications so we need to have somehow AI platforms and toolkits both for developing those applications en reaching existing applications and making people who actually need more of these applications not be limited on limited development resources that we have today now can I write source code GitHub can do I need a smart human in front of that GitHub oh absolutely that's why we as Microsoft will that's like I don't think that ever going to change it is a co-pilot we don't hand over this to be a pilot and we don't plan to and that's for the very good reasons that krina done such a brilliant job explaining you to you guys why now all of that is fantastic opportunities but it worth nothing and if you're not afraid you should be it

has to be secure and government can and hopefully will continue issuing policy and giving a frame and guard rails but let's not be fooled there are equally large amount of negative energy in the human population as much as a as a great light energy and creativity so there are dark actors everywhere in the world using the same tools that we create good things to create bad things so we have a fundamental increase in security threat which I have a couple of data points when we get to that um so this has to be part if you want your organization to be ready for AI in whatever light medium or very heavy version of that that has to be part of the conversation it has to be part of the framework it has to be part of something that you actively procure and embed in any conversation that you have because if you haven't then you're already vulnerable to everywhere out there with the same CH GPT llm model it may be dumb enough to do

my multiplications but trust me it is smart enough to create a scam campaign AI enabled applications and then finally we we will always Circle back to that killing of humans and that's I think that's a task that we have to accomplish together neither party alone can do that our education our K12 our universities can't do this alone because many many people do not study computer science for a living they study something else and they are part of that technological ecosystem that we are talking about today so how do we get from there um there are a couple of uh numbers for the data scientists here uh that's the um research we done uh internally is already public on web I'm happy to share whoever is interested in the back data we see a huge Improvement in productivity right for those of you like my friend Karen is current here around she probably needs a business case over here um we also did some some hardcore numbers where we see this productivity coming in so for

every dollar that people are do invest in AI technology and by AI technology I mean all of the stacks and not just the first layer of it um they are realizing 3.5 and 14 months now there are many people and many compan compies in the tough Financial situations so it's a very very deliberate choice to say where do I invest this money because this break even of 14 months not everybody is at Liberty to have so here is here is your choice do you go and partner with a platform that have the stuff you you don't need 285,000 gpus to train your own llm but you invest your money into eii tooling you invest your money into security you invest your money in good practice applications or do you invest your money in writing researches paper because that's what I think other people doing already for us so go use this I absolutely loved Tina's point and say let's make it actionable let's not reinvent the wheel because part of that wheel is already there

it's already rolling now let me decompose and I'm pretty sure there are a lot of scientists here so I'm not going to do depth on this one but just going to give you like because I hear a lot of this it's like this is the eii I'm talking about two things in this AI just right now for for the sake of an argument I'm talking about the natural language interface that the most of us see just like my kids and this reasoning engine that is actually powering up what we see so here is a short example of potential application so we go here and we say hey uh can you I'm asking CHT in that case um based on uh Microsoft Azure what would happen if the strings were cut now for the those of you who've done AI for some years you know how much actually energy got into that question and being able to answer that that is not trivial EI modeling here so the answer would be here if you cut the

balloons the balloons would fly away in the air now we have here an example of conversation with ch now here is the interesting thing that's fun enough it's not good enough to already hand it over to the physicist who's trying to optimized molecular chemistry guess that's too vague of an answer but it is incredible how far we've got just in the past 12 months of using those models because that wasn't possible when 3.5 get out okay so here's a conversation now this is interesting the copilot if the strings were cut the Ballon would fly up away due to the Bion force in the air a person holding the strings would lose their grip and the boxes in the background so you think of that I wasn't it took me two two attempts to see the boxes in the background in that picture would remain unaffected so it was decomposing basic physics and modeling into the picture that was by the way painted by Dolly so you can go ahead and say what if the strings

was cut what masses are connected to our strings dangerous of AI broken guitar strain now you go ask look at a minute what do you think what gas is inside of that air balloon now this this is nothing you can take from the image absolutely not this is now generic knowledge disseminated somewhere the answer is it's unlikely that this is air now here is what gbg has learned it doesn't say it is not a it it gives you the Judgment that um Katherina was talking about it is unlikely so it's all kind of giving you the human I'm not sure but I'm pretty sure some kind of things and a balloon will be floating in the air suggest that it will will be filled with a lighter gas such as helium and hydrogen okay so fast forward I'm asking now like I'm planning a party hydrogen sounds awesome um how about I use this for the barbecue you guys love right chpt I'm glad you're finding hen interesting shoot I strongly advise you against it

right so I know we still see a lot of hallucination in ches and um in EI um like Comm ation but this is really like this is a cool stuff up there right because it does know that electricity s light and all of the other things um wouldn't wouldn't actually go well with a barbecue party now this is a fun example you can run on a stage here is where it gets really interesting let's imagine today we can't cut the number of balloons it's still very judgmental those models but that we are exploring and we are developing veling this model so fast and that the pace of innovation is so incredible that in a very soon time we expect and I'm I'm not a future reader so otherwise I would do my job somewhere else um but if you can put that into technology that predicts the medical um diseases and can calculate how many cells of a particular um type are actually on the particular sample that saves us time we're still talking cop

Pilots guide let's not forget there will be still always a human at a steering wheel of that but that's what we say what is happening today and we see this coming from Microsoft developing one co-pilot or co-pilot application into a multitude of security applications sales knowledge workers the main expert but most importantly this piece this piece is now exploding for those of you who has been at a Hanover Fair um just couple past months you've seen that there is almost no industrial application it goes away without having this Y and bed technology and this is how it looks right now with many many many many hours so this is like real life examples from real life customers and that's what we talk about the age of co-pilots and we talk about the models and Frameworks that we happy to go talk either through the coffee break at any given time uh to you about because that's what it takes to get ready fori it takes all of these pillars and all of this layers just

like those of you who studied at the time when I studied ISO 8 layers these are the layers for the getting ready to Y now let me fast forward because here is a time running and I'm running out of time let me give you the point on data because I think this is interesting so this is my conversation with the I I've asked co-pilot um to analyze all of my communications and tell me a fancy story what do they think what a superhero I am and I came to the conclusion that based on my communication if I would be a superhero in the universe I would be an Iron Man now funny enough we done this as a team so we came up cing tell me like eight Iron Men I'm like okay the data on which the jgpt is trying to PR this an is not well enough so what I did then is I actually asked them to take the whole fandom Wiki into calary so I Linked In a Different data set

to the same question and AI came out and say Hey you are actually the Oracle from The Matrix which means depending on what databases and analytics you put behind you will get a different answer and this is the job for all of you here go figure it out what your organization needs because just relying on the standardized model and not putting the data behind you are not getting what you need out of chpt open AI on any of this so let me skip forward a couple of other things I I wanted to talk to you about this is just like super quick this is all a partnership neither of us nowhere regardless where are you in the world will be capable of getting their alone so we have to be partners in crimes here um and um really make it happen so let me let me close with saying here is what you really need in my opinion and my opinion only to be ready number one you need imagination a lot of it because

that's nothing that can come from machine and you need a responsible approach and if you get this two you actually can start painting the future just like Monet renoa and pisero enabled us to see the world through the different lens and with that I will open for questions thank you Berlin

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.