Realistically, what can these types of commitments look like for the AI frontier companies, moving forward?
They're releasing ever more powerful models with stronger offensive capabilities. So do they have to help bolster the defense of all existing software, just... forever?
If we advance both the offence and defense with each new release, is this sustainable?
Oh this is brilliant, I've spent the last month doing something just like this. As a challenge, no libraries allowed besides Python standard libs (so no numpy).
Started with Word2Vec, built an RNN, then LSTM and am halfway through building transformer architecture.
I reckon that in 50 years the very idea of code existing will be esoteric knowledge, a bit like binary. We simply won't care to think at that level of abstraction anymore.
I'd agree on the voice transcription; it seems so much more accurate than the other frontier models I've used. I often speak to Grok and paste the transcribed output to Claude!