Appreciate hearing from practicing researchers like yourselves in the field, very informative to hear what might be working in practice and not... I do wonder though, transformer based models may just ultimately play the supervisory role to help scale these types of breakthroughs, by knowing when to tap-in different types of models, runs lab work / simulations, and deploy diagnostic programs via tool calling, scaled by deep memory across experiment workflows?
And the "magical" part is not really for devs who have their dev env setup. For them it's an "oh so what" moment.
It's for:
- New joiners
- Non-engineers
- Devs contributing to codebases they have not set up locally (and now no longer need to setup locally!)
Only Art and Science raise humanity to the sublime. But the magic isn't in the output. It's in the process of creation and discovery, and the experience of feeling and understanding.
@dwr most people are filling the gaps with just in time software until then - scheduled supervisor agents that spin up and spawn subagents when something to chew through works well
@joshelman it seems that gap is not showing up in new opportunities in company's yet though - there's a bifurcation between engineer and product marketer with little room for a product builder?
If you have a few minutes today, watch this short film. I highly recommend it. You won’t be disappointed.
It’s one of the winners of AIF and a film that everyone I meet keeps mentioning. It’s mesmerizing, with a brilliant concept and exceptional execution.
Costa Verde. Directed by Léo Cannone & New Forest Film.