We are past chatbots in programming for a while now. It’s llms with tool calling capabilities now which work in an agentic loop. LLMs are extrapolators so the input context is important (extrapolating from missing information leads to hallucinations). With this workflow the LLM can construct its own context by using tools which leads to better results.
I haven’t met a lot of people who actually understood machine learning that say things like LLMs ‘a known scam’.
I agree that the industry, is massively overhyping the future capabilities of this kind of software in order to maintain their valuations… but the framing that AI (neural network-based machine learning) is useless is social media brain rot, not an accurate survey of the state of machine learning.
We are past chatbots in programming for a while now. It’s llms with tool calling capabilities now which work in an agentic loop. LLMs are extrapolators so the input context is important (extrapolating from missing information leads to hallucinations). With this workflow the LLM can construct its own context by using tools which leads to better results.
I haven’t met a lot of people who actually understood machine learning that say things like LLMs ‘a known scam’.
I agree that the industry, is massively overhyping the future capabilities of this kind of software in order to maintain their valuations… but the framing that AI (neural network-based machine learning) is useless is social media brain rot, not an accurate survey of the state of machine learning.