Welcome to the KuppingerCole Analyst Chat. I'm your host. My name is Matthias Reinwarth, I'm the director of the Practice Identity and Access Management here at KuppingerCole Analysts. My guest today is, for the first time, not an analyst, but one of the important people in the background of KuppingerCole. My guest today is Joerg Resch. He is one of the founders of KuppingerCole Analysts and he's the mastermind behind the EIC agenda every year. Hi, Joerg, good to see you.
Hi Matthias, I'm pleased to be your guest.
Great to have you. And we want to continue a discussion that I started with Alexei a few weeks ago. We talked about the role and the importance of generative artificial intelligence, machine learning, there is a big buzz around ChatGPT and image creation. And we want to have a kind of a different perspective on that topic. What is your first statement when it comes to talking about ChatGPT, about DALL-E, about all these generative machine learning algorithms that everybody's currently trying out?
Yeah, it's an interesting question. Thank you. ChatGPT and similar generative pre-trained transformers, that’s the acronym GPT. I think that it's more than just a software. It's a new kind of platform, or would we say, a new medium, which is now entering the way we are working, the way we are living together in a very high speed. I think there hasn't been any service application before that created so many users in such a short time. So this is something very new and it seems to be very attractive to us humans to deal with those GPTs, as for us, it is very unusual that other species or systems are able to use our language. There is nothing around aside from ourselves. Now with ChatGPT, this is changing. There are some machines who have learned using our language. And that's a great thing for us, isn't it?
It is. And if we consider ourself also to be defined from a more philosophical point of view, also being defined through language and through the way how we can communicate and that we do communicate with each other, we are adding another - you say species - another conversation partner to the conversation and to the way we are interacting with our outside world. How does it change if the other “person” is no longer a person, no longer a carbon based lifeform, but a technology that has been pre-trained. What does it mean and how does that change the way of interacting?
Interesting question. I think, I've talked to quite a few people about that and some of them admitted that they are some evenings they're maybe together with a glass of wine. They prefer talking to something like ChatGPT instead of people. So these things start to occur and happen. But aside from that, we should not compare ourselves to these kinds of machines. This is something very, very different. AIs as we know them today, those generative AIs. They are very different to humans and the way we are thinking. The human brain is the most fascinating system in this world. It contains, I think it's 90 billion neurons which are connected with many other neurons. It's extremely complex of a machine, and it is working, it’s powered by, I think, 15 or 20 watt. So it doesn't consume a lot of energy. Each neuron is an analog device, it cannot store just zeros and ones, it can store many different information. So there is an endless room of dimensions we can take advantage of, which is sometimes triggered by emotions, by hormones or by chemical substances like drugs or alcohol. So, you know, this is an extremely complex machine and it probably will never be simulated in a way that we could say, This is another human like existence, a creature. This will never be the case. So we don't talk about systems that self-reflect. They do not know who they are. We do know who we are. So we should not waste any time to compare those intelligences with what we are. But nevertheless, those are artificial intelligences, they are artificial. And intelligence, if we define it as something that can solve problems, we get into a discussion area where we can now start to understand how generative AI will change the way we work in future time and how fast this probably will happen. This is a much more important discussion than something like, you know, mathematicians like to do, whether artificial intelligence is similar to human intelligence. That doesn't play a big role.
Right, if we talk about ChatGPT where it adds value, if you look at it for a second, just as a tool. Where does, it where should it, where can it support the way we are doing business today, how we are working today. How will it influence, from your perspective, the way we will do... how we will interact in working life in, say six months from now?
Six months is a very short period of time. I've started using ChatGPT as an amplifier. I have a thought and I use ChatGPT to shorten the time from idea to concept. Still, the idea is mine, I tried to have ChatGPT as a tool, use it as a tool and try to get better using ChatGPT than not using it. And this most of the time I have to say, works out pretty well. So for me, if I create a document or a concept or maybe an Excel sheet or whatever, doing it with ChatGPT means I need less time for a better result. Still, I need to have the capability to find out whether the content generated by ChatGPT is correct, is true, is something I would put myself behind and say, that could be done by me as well. We know of the problem that ChatGPT sometimes is inventing facts, or is not disclosing sources of something. It is pretending or is trying to present it as a fact. So that sometimes needs further confirmation through other sources. But in general, of course, it's, for me it's a great tool. And it will immensely increase my ability to output concepts in future time.
Right, but what you’ve said offers or shows, a large level or a huge level of self-discipline. So it's a kind of guided content creation, controlled, managed content creation, but you are influencing that in a strong way so that it really forms the thoughts that you are aiming at. I do not expect that everybody who uses such a technology will apply that level of discipline, that level of management, that level of control, and also checking back whether the output is really what they want it to be. But you see a danger in that, that it's really just running loose?
Well, like with each technology, there is always the risk of dual use. You might’ve heard of that occurrence that, I think it wasn’t ChatGPT, it was a different generative AI that created 40,000 bad molecules that could be used to, you know, poison somebody or create a chemical weapon, artificial molecules, 40,000 in 6 hours time. So that can be used by people who should not be able to get hold of such information. So there is a problem of dual use, we have it with everything else as well, that has nothing to do with ChatGPT. And then of course there is the challenge that those language models like ChatGPT, they need to rely on huge quantities of content that is out there in the internet. And if AIs like ChatGPT start creating content themselves, this content will be fed to other AIs and those AIs then will use the AI generated content again, and there will be no instance in between verifying that content or trying to find out whether this is fake or isn’t fake. So we might see a fast deconstruction of reality and truth in the internet. That's the highest risk, I would suppose. And this is the area where we need some sort of governance, we need some sort of regulation and probably the need to mark and sign unchecked AI created content.
Yeah, I think that is an important factor, this echo chamber, so that you again and again add content that has been created through an artificial intelligence being really used as training material afterwards and then being the basis for new generation. I think this is really a threat, but a threat to this actually useful tool. If we look into the future regarding the evolution of the technology, what do you expect for ChatGPT or all these GPTs in general to happen to be the next steps? I know predictions are difficult, but what do you see, or what has been announced?
We see that, I think this year, this week on Thursday, Microsoft made an announcement, GPT-4 as the new basis that is being integrated into office 365, some additional features with a significantly larger language model. We have heard some figures flying around, but they have not been confirmed, so I would not mention them. But that new GPT-4 will be significantly larger and we can expect that it will be better and sooner or later it also will be connected to the Internet in real time. For now, GPT-3 had been trained on a fixed amount of data. I think it was some 45 terabytes of data, 175 billion words or nodes, but it didn't... you know, that was not dynamic, it was a fixed part of the Internet. Now, this will be more dynamic in future time and we can expect that the results will be much better and it will improve day by day. The usage of ChatGPT will help ChatGPT to get better every day, every second. So it will be great to see how this AI will work in half a year or one year time. Maybe with video added, with a sound added, natural sound. So that it’s not only writing, that you also can listen to a nice voice, maybe with a video behind, or whatever. So this will be changing everything we have so far experienced with the digital world. Add metaverse, add other concepts which are out there right now, then we can imagine what this will mean to our cultural lives, to our social lives, and of course, to the way how we are working. I think within the next 3 to 5 years, the way we are working will dramatically change. If I look into our own company, I know this will cause some discussion what I'll tell you now, but I think that in five years time, not even 5% of what we do today will be done by humans in future time, in five years time.
So we are at the beginning of an evolution that we cannot stop, we just need to understand it and embrace it. I think that is an important part here. And finding the right 5% that we will still be continuing to do might be the challenge there. And I think that that's really what we should be aiming at. So really embracing it, understanding it as far as we can and continuing that discussion. And that brings me to my final question for today. I've mentioned you are the mastermind of the EIC agenda, EIC, our conference in Berlin in May is coming up quite quickly. I want to continue this discussion with different standpoints regarding how we can look at this, ChatGPT and all these machine learning models and we will continue that discussion also at EIC. Maybe you can mention a few of your thoughts of how we can continue that discussion at EIC.
Oh, I think it will play a role in at least half of the sessions at EIC at some point. Also there will be ChatGPT and generative AI specific sessions, like for example, an introductory pre-conference workshop on Tuesday, May 8th with you and with Patrick Parker from EmpowerID. I have had a lot of discussions with Patrick Parker and it is amazing how he has been, how creatively he has been using ChatGPT to create some awesome results. I couldn't believe those results. And as you know, he is CEO of a software company. So the way how they use it in their software company to change the way how they produce software is just amazing. That will be a great training to get better at prompting with generative AIs like Chat CPT. That's a great session you should register for, it's invitational because we have large interest, so if you already have an EIC ticket, you should use that feature that you can create a request for invitation for that session. That's an important thing. And of course there will be a lot of things around Metaverse, about decentralized identity, decentralized infrastructures, and with all that AI powered cybersecurity and identity-first security will be in the focus of EIC 2023.
Yeah, I'm in the lucky situation that I will be there. I highly recommend to everybody that you join us there and join the discussion. Let's wait and see if this is just, as we are recording this right now in mid of March, it's just two months in the future, but two months can be long when it comes to such evolving technologies. And let's see how things have changed until then and how we can also provide insight and guardrails when it comes to that discussion around AI. In all these different types of technology areas that we are covering and in society. Thank you very much, Joerg, for being my guest today. Any final thoughts that you want to provide before we close down
Yes, just that one word: Don't fear AI, it will help you a lot and it will free our minds to care for the real big problems like global warming, like stopping war. Let those AIs do the things you don't like to do.
Perfect. So do the heavy lifting by AI and do the really interesting stuff through people. I think that is a very positive expectation and a positive outlook on that. Looking forward to continuing that discussion, meeting you in Berlin in early May for the EIC conference for anybody interested, reach out to our website and register and join us there, either virtually or in person. I would love to hear your opinions about that as well. And as always, if you have thoughts, questions, any kind of contradictions, leave them in the YouTube section below in the comments and let us know how we can cover that in a different way or in the same way and endorse us and recommend this video. Let us know how you want to continue that discussion. And I think Joerg and I are happy to receive your feedback just by mail. Reach out to us and let us know. Thank you very much, Joerg, for being my guest. Looking forward to having you very soon. We should have done this earlier. Thank you very much.
Thank you. Thank you for having me.
