Are the Foundations of AI Solid?
2023 has ushered in the age of generative AI, technology that creates documents, images, and audio in seconds. Known as Large Language Models (LLMs), this foundation has already generated lawsuits, congressional action, and billions of dollars in revenue. However, it may have deeper problems.
The core issue with LLMs is a lack of understanding, a problem that goes back to Big Data. Turing Award winner Judea Pearl says, “We live in an era that presumes Big Data to be the solution to all our problems.” However, he warns that data is profoundly dumb. Thus, we have impressive results from generative AI without context. The machine knows the following word in the sentence but cannot explain what the sentence means.
How Dumb is It
The LLM foundation is exciting for users but ultimately disappointing for the near future of AI. Due to the market response, companies such as Microsoft, Amazon, and Google are pouring billions of dollars into these technologies. However, as a technology, LLMs are little more than a clever data-mining tool.
Pearl explains, “Data can tell you that the people who took a medicine recovered faster than those who did not take it, but they can’t tell you why.” It is in knowing why that is most valuable. A simple algorithm can tell you what happened, but intelligence is required to understand why. Therefore, it is clear that the current foundations of AI are nothing more than statistics at scale.
The Next Evolution
“There is a huge space of still-unexplored application domains for AI, and a lot of the most valuable AI companies will be fundamentally new,” describes Sarah Guo of the venture firm Conviction. The success of the LLM comes as the first mainstream application of AI. Before OpenAI, most machine learning was behind the scenes or limited to research laboratories.
However, the danger of LLMs and Deep Learning is that it provides a false sense of accomplishment. Many new companies are based on using these foundational systems for specific domains — the equivalent of launching a directions app using Google Maps API.
We at Estimand believe that the next evolution will be created by the Causal Revolution pioneered by Judea Pearl. Causal AI is a new science distinguishing fact from fiction and is a foundation for artificial general intelligence (AGI). It is based on causal relations instead of statistics. As a result, this technology allows machines to understand why something happens.
In the linking of cause and effect, a machine builds context. Once an AI has context, its answers have a depth of knowledge. Instead of regurgitating something a journalist wrote in a slightly different form, the Causal AI explains the factors leading up to the event in question.
Why It Matters
There is value in LLMs and Deep Learning. However, these technologies will not get us to the next rung on the AI ladder. Instead, they are currently serving as hindrances to novel approaches. When many companies launch to serve some niche using existing technology, it crowds the marketplace with inferior solutions that distract from truly revolutionary offerings.
The second problem is that venture firms are currently focused on the people leaving OpenAI, Cohere, Anthropic, and Mistral as those who will create the future of AI. The future of AI will be made on the fringes by smart people without the baggage of confining technology. At Estimand, we come from hedge funds, IBM, and academia with a new approach proven to bring context to AI. One that is available right now.
About Estimand
Founded in April 2023, Estimand Inc. is a Delaware Corporation with principal offices in Raleigh, North Carolina. We began as a paper on causal AI that evolved into the first company to instantly reveal the factors behind global financial risk. You can learn more about Estimand by visiting our website: https://estimand.ai.
References
- Gynbaum, M. & Mac, R. (2023) The Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work. The New York Times. https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html
- Hammond, G. (2023) Big Tech is spending more than VC firms on AI startups. Ars Technica. https://arstechnica.com/ai/2023/12/big-tech-is-spending-more-than-vc-firms-on-ai-startups/
- Pearl, J. (2018) The Book of Why. Basic Books
