Watched podcasts with
Ilya Sutskever and
Andrej Karpathy over the weekend:
— Ilya Sutskever on human-like AI, philosophy, evolution, and the next era of research
— Andrej Karpathy on the end of coding, progress in AI agents, and how modern AI R&D is evolving
My personal takeaway: we need to tone down the global AI panic a bit.
- Too much of the conversation is still stuck between hype and fear.
- Too many scary predictions
- Not enough focus on practical use & adoption.
For now, the biggest shift is much more grounded:
AI is changing how workflows are built — moving them from human-first to agent-first.
At the same time, it is becoming clearer that AGI may require far more research than some of the biggest AI labs were implying to investors. And honestly, I see that as a good thing.
It gives us more time to align AI with human interests, make it more useful in real-world contexts, and keep the focus on serving sentient humans rather than feeding abstract narratives.
The opportunity in front of us is not just building more powerful models.
It is building AI systems that are practical, aligned, and actually improve how people work and live.
