Expanding intelligence

Artificial intelligence (primarily via science fiction) sparked my interest in technology as a child. Hence, much of my work this far has been towards “expanding intelligence."

This includes building better datasets for smarter financial models at Two Sigma, smarter algorithms for advertising at Snapchat, smarter anti-fraud systems at OpenSea, developer tooling for smarter video game characters at Inworld AI and new products for making learners smarter at Sizzle AI.

Back when GPT-3 was first announced in 2020, I was eager to explore working with intelligence in a more general-purpose form.

I secured early private beta access to the model for a company hackathon at Spatial and built "Geoffrey" — an AI agent that could intelligently populate a 3D workspace with relevant files and media retrieved from the internet and across a user's Google Drive, Figma, Slack and Miro accounts. Although the technology was young and had its kinks, the week-long project gave me a taste of what the next era of human-computer interaction could evolve to be.

Not until the launch of ChatGPT in late 2022 did the entire world start to shift, with capital, compute, research and mindshare all gravitating towards the field at an unprecedented pace. Over time, I became increasingly convinced that a close loop between the development of these frontier models and their real-world application would maximize the potential for their capability.

Thus, I’m joining OpenAI to build [a new product] that will leverage the latest of its research.

This is admittedly a big change for me — it ends my streak of early stage startups and also begins my new next chapter in San Francisco. But after chatting more with Sam Altman about [the product], I became so infected by his enthusiasm that I decided to hop aboard nonetheless.

Endless opportunity awaits a future where the intelligence of humans & machines are greatly expanded and the relationship between them is tightened, so I look forward to playing a part in pushing the frontier of possibility.

Onwards to a safe AGI 🫡