What is generative AI?
Björn: Generative AI is quite important, right? Before we get to that, what is Generative AI, from your perspective? Bent: Generative AI is naturally the new guy on the block, if you will, but it’s not new, as we have been working with machine learning using neural networks for a long time, and the issue has been to lift it to the next level. In 2017, Google researchers were able to bring things that seemingly were not very contextually related to become contextually related. This neural network architecture was called Transformer, hence the “T” in “GPT”. Business opportunities from generative AI
Björn: How do you see generative AI being so versatile in the usage to transform businesses? Bent: It has a huge potential. We have done a paper for the Danish industry on this, and using the sources as well. Industries that will be affected
Bent: We can see that [generative AI] actually will affect about 80% of all jobs, which is quite a lot. And then for those jobs, the impact will be quite different depending on the job. So whether you are in legal, HR, or finance, all of these different aspects are something that it has a basic knowledge about, and it’s able to converse with you on these topics. And that also goes across industries. For natural reasons, the industries that are really impacted are the service industries mainly. Not so much when it comes to construction, like those doing the drilling, but the service people, such as a receptionist, etc. Mainly the people that are doing what we call white collar work. What will happen when clients roll out generative AI
Björn: What are the specific use cases you see? Bent: I think when we started helping our clients roll out generative AI, we were also actually looking for specific use cases and saying, okay, this is a use case. Is there a business case for this? Maybe we can do it. But actually, that’s using the old way of looking at this technology. Björn: So rather than help the client use it for certain things, it’s more, how do we deliver a regulated model that is secure for the client, so they can, without any fear, start to experiment themselves? Bent: Exactly. So a lot of ideas can come up from the ground, and it’s very much about ensuring that you are supporting this also as a company. Normalizing the use in the office
Bent: Recently, people called into a radio station wondering if they should tell their employer that they’re using ChatGPT? So it should not be a hidden thing, it should be promoted. Is there fear?
Björn: Is it the governance or is it maybe the fear of people being obsolete now with this revolution? Bent: Yeah, it’s not about becoming obsolete. It’s a lot about regulation. So there’s a misconception that this is extremely dangerous, which is not the case, but that takes a lot of time and that’s the hindrance right now. And then the belief, because people don’t know how it works, is also a problem. Framework and regulation
Björn: What would be your best recommendation on how to get started? Bent: Step one is to know what [generative AI] is. What is it able to do and what is it not that good at? Get some knowledge. Second point, create a framework where you can have your employees playing around with this. And make sure that this fits your requirements from a regulatory perspective as well. And then the third one is, begin to teach not only about [generative AI], but especially about the whole prompting. This thing about talking to a model and talking to data is completely new for everyone.
