> Anything but the simplest tooling is not transferable between model generations, let alone completely different families.
It is transferable-yes, you will get issues if you take prompts and workflows tuned for one model and send them to another unchanged. But, most of the time, fixing it is just tinkering with some prompt templates
People port solutions between models all the time. It takes some work, but the amount of work involved is tractable
Plus: this is absolutely the kind of task a coding agent can accelerate
The biggest risk is if your solution is at the frontier of capability, and a competing model (even another frontier model) just can’t do it. But a lot of use cases, that isn’t the case. And even if that is the case today, decent odds in a few more months it won’t be
If you're talking about APIs and SDKs, whether direct API calls or driving tools like Claude code or codex with human out of the loop, I think that's actually fairly straightforward to switch between the various tools.
If you're talking about output quality, then yeah, that's not as easy. But for product outputs (building a customer service agent or something like that), having a well-designed eval harness and doing testing and iteration can get you some degree of convergence between the models of similar generations. Coding is similar (iterate, measure), but less easy to eval.
It's not that hard to make it generic. It does take a little work, but really it boils down to figuring out how to make things work with the "dumbest" model in your set.
It never made sense to me why cars and pedestrians need to share the same spaces. Why can't we have more efficient walking routes that are away from cars?
if you have roads shared with pedestrians and cars (and bikes!) you can build denser cities.
I lived real downtown in Tokyo and my street was like "1.5" lanes wide (if cars were coming in both directions one basically needs to pull over and stop). I could just walk in the middle of the street. There was no sidewalk. No street parking of course. Cars would drive down at 15km/h or whatever, and slow to a crawl if people were in the street.
Straight lines are efficient walking routes, and ... well... that might involve just crossing the street directly! Every layer of grade separation gets in the way of that.
End result of all of this is less pavement to maintain, slower drivers (-> safer!), good walking and cycling conditions, etc etc etc.
In the same way that numeracy skills were in the blast radius of the Colossus.
People seriously underestimate how underpowered and tiny llms are for the tasks they need to solve.
A trillion parameter model can't tell the difference between left and right. We will need to grow them millions to trillions of times before they are half as good as AI boosters claim they are.
This isn't the end of thinking any more than the watt steam engine was the end of horses. It will be centuries before we get there. And by that point the difference between man and machine will be at best academic.
And where I live you can sometimes still see a horse-drawn cart on the roads. Albeit rarely - they are forbidden. 10 or 20 years ago it was more common. I rode one once a bunch of years ago, went with its owner to harvest construction materials from a collapsed building.
Plus, on the way here today, I passed over a favela-type assemblage I can only describe as festering. No offense meant to its inhabitants - they are more real human beings than any of us.
Point being, the future has been here for long enough. And, as the adage goes, it's not evenly distributed. I never made any money from being strong or being intelligent; my usefulness, like that of any "nerd", was rooted in knowledge arbitrage. Now that the psychopaths in charge have cognitive prosthetics, I expect to be culled.
We've got army block, FSB block, technocrats, bureaucrats and oligarchs. The usual (more or less) story.
The real problem is - we don't have system that scales horizontally. So when Putin goes people will have to deal with the vertical system he created for himself.
The problem here is this "for himself" part.
For this system to work you will have to be a new Putin (at least for some time) and for this you will have to enforce your decisions and shape your new system. Top to bottom.
Best thing that can happen to Russian (realistically) is that the power will be given to technocrats.
They are not neccesarily more liberal, but they have real education, they do understand a thing or to about economics, open borders, sharing of knowledge etc.
They won't be able to quickly change Russia, but given some time they can reshape it step by step.
Alas - we have FSB and Army blocks, high level of corruption and millions of people who see people like Putin as the best choice. They don't need progress and responsibility. They need their empire back even if they are just peasants with serfdom included.
In the 80s, 90s and 00s that's what they thought about coding.
Then when the salaries got good every pretended to have always been a nerd and really into everything nerd. With the result that they kicked all the nerds out.
If you consider what assemblers and compilers do programming, sure.
But men didn't kick them out, technology did. Von Numan famously forbid the Eniac from ever being used for assembly when you had a perfectly cheap secretary pool to do the assembly by hand.
Low creativity repetitive work requiring great attention to detail is what the early female programmers did and what was automated first.
If we ever get deterministic AI the same will happen up the chain. I'm not holding my breath for the current generation of models, or the upcoming ones I've seen in papers.
That's underselling their role. One of those ladies doing the assembling for Von Numan was Grace Hopper, who then used that expertise to develop the first compilers.
reply