Progress in the applied domain (the sort of progress that makes a different in the economy) will come predominantly from integrating and orchestrating LLMs, with improvements to models adding a little bit of extra fuel on top.
If we never get any model better than what we have now (several GPT-4-quality models and some stronger models like o1/o3) we will still have at least a decade of improvements and growth across the entire economy and society.
We haven't even scratched the surface in the quest to understand how to best integrate and orchestrate LLMs effectively. These are very early days. There's still tons of work to do in memory, RAG, tool calling, agentic workflows, UI/UX, QA, security, ...
At this time, not more than 0.01% of the applications and services that can be built using currently available AI and that can meaningfully increase productivity and quality have been built or even planned.
We may or may not get to AGI/ASI soon with the current stack (I'm actually cautiously optimistic), but the obsessive jump from the latest research progress at the frontier labs to applied AI effectiveness is misguided.
That's a nice euphemism for "imminent mass layoffs and a race to the bottom"...
The losers here are algorithm junkies who refuse to learn new skills and want to solve yesterday’s problems.
Ain't gonna change unless we genetically engineer our dumbass evolutionary history out of ourselves.
Up to you to figure out which will hold.
We are accelerating by preferring short-term gains. Like a fire becoming an explosion, that's modern society. Corps now throw the future under the bus for a slight boost in short-term value.
The industrial revolution is not an apt analogy. Humans were still too essential to getting factories to actually work. Horses becoming useless - by no fault of their own - is an apt analogy. We are rushing to a world where humans can be fully replaced.
This "humans will always be in the loop no matter what" is just Cope. We simply don't know what will happen or the what the upper bound of AI capabilities will be. But 100% automation and humans as knowledge workers being as useless vs AI as horses vs cars is no longer sci-fi. We don't know if it will, but this is a future that actually could happen within our lifetimes.
Were we at a peak 1000+ years ago and have only gone downhill since at every technological breakthrough?
ChatGPT also seems to also be building a history of my queries and my prompts are getting shorter and shorter because it already knows my frameworks, databases, operating system, and common problems I’m solving
What I’ve learned is that for the most part that AI revolution is not going to be because of PHD-level LLMs. It will be because people are better equipped to use the high-schooler level LLMs to do their work more efficiently.
We have some knowledge graph experiments where LLMs continuously monitor user actions on Slack, GitHub etc and build up an expertise store. It learns about your work, your workflows and then you can RAG them.
In user testing, people most closely associated this experience to having someone just being able to read their minds and essentially auto-suggest their work outputs. Basically it’s like another team member.
Since these are just nodes in a knowledge graph, you can mix and match expertise bases that span several skills too. Eg: A Pm who understands the nuances of technical feasibility.
And it didn’t require user training or prompting LLMs.
So while GPT-5 may be delayed, I don’t think that’s stopping or slowing down a revolution in knowledge-worker productivity.