Anna Gressel: Yeah, it's true. We've all seen technology hype cycles before, so it's not an unreasonable point to raise.
Katherine Forrest: Well, indulge me for a minute.
Anna Gressel: I can hear in your voice you have this morning caffeine thing going. So, you take it away, Katherine.
Katherine Forrest: I do. I do. I've got that morning caffeine thing. So here is how I have come to think about comments that people make about AI being in some sort of hype cycle, which is that people treat AI as if it's monolithic. And when generative AI first became part of public awareness in the late fall of 2022, or really when Kevin Roose from the New York Times had his famous article about his Valentine's conversations, with the chatbot in February of 2023, the world really woke up to this new technology.
Anna Gressel: Yeah, we all started to wonder how fast it would change things.
Katherine Forrest: Exactly.
Anna Gressel: And you and I started fielding calls from clients who wanted to know what they should be doing in the area.
Katherine Forrest: Yeah, there was this sense that AI would have a temporally and technically similar impact across all industries and our civil society all at the same time, as if suddenly generative AI were going to make the same impact on everyone in somehow the same way at the same time. And that's just not the way it was.
Anna Gressel: Yeah, no one really knew what the impact would be at that time either.
Katherine Forrest: And some really still don't. And as I say, we're at the beginning of the beginning of the beginning of all this.
Anna Gressel: But we have actually seen some pretty incredible use cases actively deployed in a number of industries.
Katherine Forrest: Precisely. And what we have seen is that AI, and particularly generative AI, has a number of what can absolutely be called transformative use cases, but it impacts different industries differently.
Anna Gressel: Yeah, I mean, it is this beginning of the beginning that you've been talking about, right?
Katherine Forrest: Right, and so let's talk about some of the areas where we've seen some particularly transformative use cases.
Anna Gressel: Without giving away any of the secret sauce.
Katherine Forrest: We never give away our secrets or the secrets of our clients. But coding we can say—computer coding is one of the first areas that comes to mind where there's been a true transformation. Any business that is heavily dependent on writing new code, new code for products or code to solve particular issues within existing products, AI is now responsible for more than 60% of all new coding in the United States.
Anna Gressel: And businesses that develop in-house software tools are at their core seeing some of the real transformative potential of AI.
Katherine Forrest: And then I would add in another set of use cases generally, and that's in the area of financial services and asset management, because there, there are a number of ways in which AI and machine learning, which have been used for years in trading in these sectors, they have actually gotten comfortable with adopting generative AI tools in a number of functions, including some additional customer service uses, market intelligence, deal diligence, internal analyses for credit agreements or financials among others. So, we're really seeing a lot of change in that area.
Anna Gressel: Agreed, Katherine, and that's just to name a few examples. But let me add in pharma as another area that's really adopted and is experiencing some of the benefits of GenAI's potential. We already know there's drug development that's going on and is being enhanced by GenAI, new molecules discovered, and important information revealed on things like toxicity. And you know my favorite thing ever is protein folding for folks who know me. That's like a true interest of mine.
Katherine Forrest: We’re going to one day count how many times you mentioned protein folding in a podcast about AI. Anyway, absolutely. And the medical field generally is another area that's seeing real transformation. There are a number of GenAI tools that are being deployed for diagnostics and testing in a variety of ways.
Anna Gressel: And we can also turn to manufacturing and supply chain use cases. So those are places that are also seeing GenAI with real impact tools that are able to assist with inventory management, maintenance, prediction, as well as on the line itself.
Katherine Forrest: Okay, and then I want to throw in, of course, one of the most transformative uses of all, which is the fourth graders' homework. GenAI has a real transformative potential that lot of fourth graders have found.
Anna Gressel: And not without controversy for those fourth graders and all of their teachers. But we can certainly say that GenAI tools are able to draft narrative text and they can save time for businesses and automate all kinds of simple daily tasks.
Katherine Forrest: And how about in our legal area? I would say that we're only now just seeing GenAI being deployed into the legal area. There have been a lot of tools that we've been talking about for really over a year now and seeing a velocity of change in those tools, but it's really just coming now into active use in this area.
Anna Gressel: Yeah, I wouldn't necessarily call legal one of the top adopters in GenAI because accuracy is still such a big issue and there are still hallucinations we're seeing in the tools, so that can be a real barrier to adoption.
Katherine Forrest: But we are actually, I think, this fall of 2024, seeing the adoption and now use in law firms. And I think we're going to see more and more of it as we head into ‘25. And it's tools that assist with sort of specific tasks, like tools that are assisting with contract drafting, that can summarize large volumes of documents.
Anna Gressel: And there are tools that can write chronologies or outlines.
Katherine Forrest: Or come up with ideas to respond to arguments raised in a complaint or a brief. But I think that we're probably both aligned in predicting that these tools are going to be even more transformative in 2025 in the legal area.
Anna Gressel: Yeah, I would completely agree with that. And let me just name some of the model developments that we also saw that are supporting the velocity of change in this area. We started with narrow AI tools, then we had LLMs, large language models, then MLLMs, multimodal large language models, or just multimodal models. There are now all kinds of fine-tuned models, there are mixture of expert models.
Katherine Forrest: Agentic AI.
Anna Gressel: Yeah, agentic AI is huge. And we're also going to see in the coming year adoption of more compound AI systems comprised of multiple models.
Katherine Forrest: Right, and now we've got these things called smaller models and larger models and we'll talk about those and the differences in subsequent episodes that are geared towards really different use cases and...
Anna Gressel: Yeah, and I mean, think there's been an unbelievable set of advances and capabilities just in terms of the new Llama herd of models, the Falcon 2 Mamba-based model, the OpenAI o1 model.
Katherine Forrest: Along with various advances in the Claude models and the Gemini models.
Anna Gressel: Well, I think there's been so much we have said and could say about transformative AI that, you know, we're now out of time. But in some of our next few episodes, we'll talk about some of the reasoning capabilities of some of these models. And I, for one, am convinced we're in one of the most transformative moments in terms of the actual technical developments happening in this area.
Katherine Forrest: And I'm right there with you. And this is Katherine Forrest.
Anna Gressel: And Anna Gressel.
Katherine Forrest: And thanks for joining us for today's episode of “Waking Up With AI.”
