Let’s face it: after-the-fact API guardrails are not the right safety tool for frontier models.
They don’t make dangerous capabilities disappear. They just hide them behind a brittle interface that can be easily jailbroken.
A better safety agenda:
- don’t train models for very high-risk capabilities without strong evals, justification, and containment
- use staged release, as pioneered by @IreneSolaiman, from trusted testers to broader access, and open release for transparency and accountability
- massively support open-source AI so the gap between players does not become so large that a few closed labs and governments end up with overwhelming capabilities and power over everyone else
- enable independent evaluation instead of asking everyone to trust a black-box API
- give law enforcement, courts, regulators, auditors, journalists, and civil society strong AI tools to detect, investigate, and hold accountable unlawful uses of AI
Safety means transparency, staged deployment, distributed power, and making sure democratic institutions can actually enforce the law.
@perreaux Most annoying is that noone is covering the accessibility angle, notably as the McGill/Maisonneuve one has the only elevator down to Green line. Signed, Dad with a stroller who wants fresh air and not the Centre Eaton
Creativity is deceptively hard, because it’s our inexperienced reflex to want to be “great”, that in itself produces bad:
Wanting to create a magnum opus creates things that have too many elements and are overcooked. The first song I ever produced was the most complex I ever did. I thought it was epic at the time but it was terrible looking back.
Wanting to create something wildly original goes too far and loses subtlety and nuance. In my intermediate music production phase I made very strange and original music that had no hooks or real listening pleasure.
The thing that makes us junior, is striving for a caricature version of greatness. The sooner you can destroy that caricature, the sooner your taste and abilities can mature.
@gregisenberg@sebpaquet Thanks for shining a light on this. The picture you paint mostly shows one thing - you haven’t found the right attorney for your mindset. The models are changing, and like any profession, where value is created is the only thing that matters.