Despite what seemed like explosive market growth for AI during 2023, challenges remain for the technology, and many enterprises have yet to progress from proof-of-concept projects to commercial deployments. But, they are challenges an entire industry is working to solve.
Two of the top analysts studying AI–Leonard Lee, executive analyst and founder of neXt Curve, and Karl Freund, founder and principal analyst at Cambrian-AI Research–said as much during last week’s neXt Curve reThink Webcast “Setting the Stage for AI in 2024.”
Speaking about 2023’s AI boom, which triggered a spike in the stock market for fortunes of Nvidia, among others, Freund said during the chat, “[Last year] was the first time in my career I've seen a multi-billion-dollar market spring up out of nothing. I have been an AI analyst for eight years watching the technology kind of progress, and now suddenly, hockey sticks.”
Lee added, “Despite what folks might think, this technology still suffers from a lot of deficiencies, whether it's model drift or hallucinations or model collapse, there are a lot of things that have not been solved… There are methods surfacing to mitigate some of these undesirable side effects, but it is still a learning curve.”
That is why AI, for all the headlines and AI chip news of the last year, remains a technology that many enterprises are still just testing. Lee said enterprises are “carefully assessing whether or not these generative AI technologies that are available today are suitable for anything that's business critical–and beneath or above business critical is mission, critical, safety critical, all these things that require, like, an exorbitant number of nines of reliability… They're being very measured about where they apply generative AI or AI in general. Because it is probabilistic, and a lot of these applications require… deterministic capabilities.”
Because of this, Lee said companies may continue to be “quick to experiment” but “slow to adopt.” He added, “Like with IoT over the years, there are tons of PoCs, but little actual deployment right now.”
Freund suggested that the AI ecosystem will solve technology problems in time, but that companies working with the technology throughout this year need to focus on what it can do now, not what they wish it could do. “There's lots of stuff AI could do, but it's not going to be able to do in 2024. Focus on what it can do in 2024, while minimizing the risks” of working with AI technology, he said.