Intelligence, Human and Artificial
The most recent meeting of the Austin CTO Club featured an extremely vigorous discussion under the theme AI Hype versus Reality. Listening this discussion led me to a new AI strategy.
Everyone in the meeting is exploring use cases for AI. And everyone is working hard to keep up with the rapid pace of development in AI tooling. Very few among us need genuinely bleeding-edge capabilities today. Yet we all recognize that yesterdays bleeding edge becomes tomorrow’s new normal. What strategy makes sense in response to these circumstances?
Many participants are significantly more advanced than I am in their use of models and harnesses. Given the extremely broad variation in application of these tools, I don’t think anyone can stay abreast of all of the developments. We are naturally specializing our understanding around our specific use cases. And so I’m hardly alone in having very limited capability compared with the group.
Not only are models improving fast, but harnesses that wrap them and techniques for using them continue to evolve at a similar pace. There are too many margins for any individual to cover, even though we have to place bets as we go.
My new strategy is two-pronged. First, use the tools daily, engaging in direct exploration of their evolving capabilities in the context of my use cases. Second, borrow aggressively from others in the group as they explore margins that remain out of scope for me. Only through such a group can one hope to touch enough margins to keep up broadly.
We must leverage our collective human intelligence to understand AI intelligence.