THE FUTURE OF AI agents operate across extended time horizons, managing complex workflows without continuous human oversight. For many observers, such autonomous operation crosses traditional AGI thresholds that emphasised human-level performance across diverse tasks.
“ The coding agent we just released has been one of my biggest AI moments,” Sam says.“ You can give it a bunch of tasks, it goes and works in the background, it’ s really quite smart, it can handle long chains of things, and then you get to just sit there and say‘ yes to this one, no to that one, try again’.”
Today’ s models might operate with limited autonomy periods, but Sam suggests the foundations are there for extended independent operation.“ Maybe today it’ s like an intern that can work for a couple of hours, but at some point it’ ll be like an experienced engineer working for days,” he explains.
Ultimately, development could extend into research applications where AI systems could operate autonomously on complex problems.“ I expect next year that in some limited cases, we will start to see agents that can help us discover new knowledge or figure out solutions to business problems,” he says.“ At some point you’ re going to have an AI scientist that can go discover new science, and that will be a significant moment in the world.”
Looking ahead, the models scheduled for release within 24 months will enable businesses to delegate their most complex problems to AI systems with confidence in autonomous resolution.
“ The rate of progress is truly astonishing... The actual moment doesn ' t really matter a whole lot”
SRIDHAR RAMASWAMY, CEO, SNOWFLAKE
“ I think we’ ll be at the point next year where you can not only use a system to automate business processes or build some policy services, but you can really say‘ I have this hugely important problem in my business. I will throw a ton of compute at it if you can solve it,’ and the models will be able to figure out things that teams of people on their own can’ t do,” Sam explains.
These capabilities will span industryspecific challenges that currently require substantial human expertise, including semiconductor design optimisation and biotechnology research acceleration.“ If you’ re a chip design company, say‘ Go design me a better chip than I could have possibly designed before.’ If you’ re a biotech company trying to cure some disease, just say‘ Go work on this for me.’ That’ s not so far away,” Sam notes.
aimagazine. com 61