(Devonyu/iStockphoto/Getty Images)
AI is the buzzword du jour. The emerging industry has kept investors optimistic and stirred up market euphoria throughout 2023.
Of all S&P 500 companies, 152 cited the term "AI" during their earnings call for the third quarter, according to FactSet data. That's the second-highest number of S&P 500 companies citing "AI" on earnings calls going back at least 10 years, when FactSet began tracking the data. It trails only the previous quarter's 180 citations.
Nvidia, the AI-adjacent chipmaker, meanwhile, was the highest performing US stock this year, up 235% so far.
But will the buzz continue into 2024?
Before the Bell spoke with Marco Argenti, chief information officer at Goldman Sachs, about what comes next for AI.
Agrenti has a history of predicting what comes next in technology. Prior to joining Goldman Sachs, he served as vice president of technology of Amazon Web Services, where he started and ran several AWS businesses, including mobile, serverless computing, Internet of Things and augmented and virtual reality.
This interview has been edited for length and clarity.
What are some of the macro trends you're seeing for AI next year?
My first prediction is what I call "the emergence of hybrid AI." The pendulum is going to swing and land in the middle of larger foundational models like ChatGPT and Gemini and the use of smaller, fine-tuned AI models. Those actors are going to cooperate. We've discovered at Goldman that you can really use the best of both in close coordination.
So the large foundational models have extraordinary reasoning capabilities. They can orchestrate where to go and reach out for data. The smaller models have the advantage of being able to run on a more constrained infrastructure environment that can be fine tuned for data privacy. The larger model will process the input from the user, break it down into tasks, interact with specialized models that reside either on premises or on a virtual private cloud and then use the summarization capabilities to come up with an answer, and that's what I call "hybrid AI."
What are some companies or industries that benefit from this shift to more refined AI models?
In financial services, there are very specialized and complex use cases, for example, in the way you interpret legal documents that are behind derivative contracts. You really need to understand the language of legal finance. You can fine-tune an AI model to interpret these derivative contracts and translate them into a machine readable format for a generic AI model to use.
You can create specific models of AI that calculate optimal portfolios or asset allocation. Or that go through earnings calls and corporate filings and extract salient points.
So if a client asks a complex question like "can you tell me about all the companies that have reported more than 10% growth in a particular sector because of a particular trend?" the larger AI model could interpret that question, break it down into instructions that then go to the specialized models. They take that information, put it together and then send it back as a summary. It's like having managers and specialists working together in close coordination. I think this pattern is emerging more and is something that companies will gravitate towards.
A lot of large corporations and financial institutions have banned the use of AI. Could this specialization shift the needle on that?
This leads to my second prediction, which is going from the potential of AI to the execution of AI in a safe way. From promises to results.
I think 2024 will bring a lot of these proof of concepts to life and provide a return on investment that will actually make an impact on organizations. It will increase the productivity of developers and operations, or offer a better level of customer support. It will streamline operations. There will also be a very strong focus on safety and governance, which are less of a concern when you're testing the technology.
There have been a lot of discussions around how to regulate AI. Nobody fully understands the potential of AI, we're mostly predicting what that could be, and nobody has full evidence of where it can go. My view is that some of this uncertainty became fear, in terms of AI being a threat to humanity. In 2024 and beyond, there's going to be a thoughtful balance to regulation that prioritizes security but also enables the ability to innovate.
We can't create barriers that put the United States, or other countries, at a disadvantage if the regulation is too restricting.
How do you think the 2024 elections will factor in to regulation and fear mongering over AI?
This will need to be front and center in the agenda. This will be a major topic because AI has already had an impact on people's lives. There are a lot of questions around the protection of intellectual property. That's something that potentially involves millions of people, there are a lot of content creators out there that don't want to be replaced and who want to be paid.
Generally, fear is fueled by a lack of understanding — not knowing or not having tried something. But hundreds of millions of people are using either ChatGPT or other forms of AI on a daily basis. As you get used to these tools, first in the consumer environment and then in the enterprise environment, and the unknowns become more familiar, those fears will ease.
We're clearly very early in the process, and the disrupting potential of this technology is something that we don't yet understand. And some will wave that flag.
AI was such a buzzword this year. Do you think that investors will become more discerning in 2024 about investing in the industry?
My final prediction is that there will be a shift in venture capital fund allocation with regards to AI. At the beginning we stuck an AI label on a lot of things.
There's going to be a shift in funding from the foundational side, people investing in creating new foundational models, which are extremely capital intensive, into the application layer. We're just seeing the beginning of investing in business-to-business and vertical applications attached to those foundational models. It's a new world and I think there's a huge market and a huge opportunity there.
It's been a tough few years for M&A and IPOs. Could AI change that?
If you look at the amount of new startups actually coming up and the number of projects, there are thousands of new companies being created. And of course, some of them are going to fail. But some of them are going to be acquired and some of them are going to go public. So I think the top of the funnel right now is quite healthy. That's a leading indicator.
Comments
Post a Comment