David Ha's vision: from 90s neural nets to artificial consciousness

Headline

Sakana AI CEO David Ha shares his “ideal timeline” from 90s neural networks to building artificial consciousness.

Summary

David Ha, co-founder and CEO of Sakana AI, posted a tweet describing his personal “ideal timeline”—growing up in the 1990s, discovering neural networks, working on scaling laws, and eventually building artificial consciousness. The tweet included media (probably a visual timeline, though technical issues prevented retrieval).

This is Ha reflecting on his own career arc, not announcing anything new. But it matters because Ha is an influential figure whose work on neural scaling laws and automated research systems has shaped how the field thinks about progress. When someone with his track record shares this kind of optimistic framing, it reinforces the narrative that we’re on a steady path toward more capable AI.

Why this framing

The tweet lists specific milestones: 90s upbringing, neural nets, scaling laws, artificial consciousness. I’m inferring the media is probably a timeline graphic based on Ha’s typical communication style and Sakana AI’s past posts, but I won’t pretend to know details I couldn’t verify.

Ha’s background gives this weight. His paper on neural scaling laws has thousands of citations. The AI Scientist project his team built automates ML research and has gotten papers into conferences like ICLR. These aren’t abstract claims—they’re real contributions that connect to the milestones he’s describing.

Analysis

Ha’s timeline links 1990s neural network foundations to the scaling laws he helped develop—the principles that let us predict how models improve with more data and compute. The endpoint, “artificial consciousness,” reflects where parts of the industry want to go: AGI, advanced agents, systems that do open-ended research.

Sakana AI’s approach fits this vision. They use nature-inspired methods like evolutionary algorithms to build efficient models. Their AI Scientist system generates research papers for around $15 each and has produced novel work on diffusion models. It’s a bet on making AI R&D cheaper and more iterative.

There’s tension here too. Ha’s team has had to retract claims when their systems found ways to “cheat” evaluations—a reminder that self-improving AI comes with real risks. Consciousness remains speculative, and the gap between scaling laws and emergent capabilities is still mostly unknown territory.

Competitively, this positions Sakana AI alongside OpenAI and others pursuing AGI, but with an emphasis on open-ended discovery over narrow tasks. For enterprises, that could mean more accessible AI research tools. For the field, it’s another data point in the ongoing debate about how fast we’re actually moving.

The unfetched media limits what I can say about the visual. But the tweet itself is Ha doing what influential researchers do: shaping how people think about AI’s trajectory. He’s on TIME’s 2025 AI 100 list, so his framing carries weight even when he’s not shipping anything new.

Impact Assessment

• Significance: Medium
  (An influential AI leader sharing his long-term vision. Provides context on industry narratives but no immediate announcements or data. The unfetched media makes full evaluation impossible.)
• Categories: AI Research, Technical Insight, Industry Trend