If I understand what you're saying, you're asking to be able to perform a very specific search query (also known, at the technical level, as a 'complex' search query) such that it also returns to you "the complete list" of what you looking for.
Will Bryk, the CEO of Exa-dot-ai, seems to be putting together such a "search solution" (AI-bas…
If I understand what you're saying, you're asking to be able to perform a very specific search query (also known, at the technical level, as a 'complex' search query) such that it also returns to you "the complete list" of what you looking for.
Will Bryk, the CEO of Exa-dot-ai, seems to be putting together such a "search solution" (AI-based) for companies to then integrate into their platforms to then offer them to their customers.
So, a potential customer for Exa could be the Substack platform.
More on Will Bryk's vision here (link below). An example of a very specify search query he gives is (paraphrasing): "[I'm an in-house researcher at a Venture Capitalist firm. I need to help the startups I fund hire PhDs who specialize in 'AI Chatbots']: return to me a list of all PhDs who have written a thesis on 'AI chatbots'." Result: it returns the complete list.
Most heuristic AI is search without a global optimimum; so good not perfect answers given the space being searched. Relational Query and Search Capabilities are increasingly overlapping and looking like each other. There are also newer technologies that can emulate search like Tsetlin machines that have not mainstreamed yet. LLMs with transformer architecture (ChatGPT and buddies) are computationally inefficient meaning you pay for tokens in memory and compute usage costs that are pretty high.
If I understand what you're saying, you're asking to be able to perform a very specific search query (also known, at the technical level, as a 'complex' search query) such that it also returns to you "the complete list" of what you looking for.
Will Bryk, the CEO of Exa-dot-ai, seems to be putting together such a "search solution" (AI-based) for companies to then integrate into their platforms to then offer them to their customers.
So, a potential customer for Exa could be the Substack platform.
More on Will Bryk's vision here (link below). An example of a very specify search query he gives is (paraphrasing): "[I'm an in-house researcher at a Venture Capitalist firm. I need to help the startups I fund hire PhDs who specialize in 'AI Chatbots']: return to me a list of all PhDs who have written a thesis on 'AI chatbots'." Result: it returns the complete list.
https://www.latent.space/p/exa
Most heuristic AI is search without a global optimimum; so good not perfect answers given the space being searched. Relational Query and Search Capabilities are increasingly overlapping and looking like each other. There are also newer technologies that can emulate search like Tsetlin machines that have not mainstreamed yet. LLMs with transformer architecture (ChatGPT and buddies) are computationally inefficient meaning you pay for tokens in memory and compute usage costs that are pretty high.