10x Lies on Productivity

I had a friend come into town recently, and over a little bit of whisky, we started discussing AI's capabilities.  My friend definitely falls within the technical realm - he's a professor of engineering and regularly needs to create small programs to use in his research projects, but he's not enmeshed in the bizarre world of software engineering.

Still, I was taken aback when he asked me about my use of coding agents.  "So, do you feel like you're 9 or 10 times more productive than you were before?"  The question almost made me spit my whisky out.

After a little bit of contemplation, I gave him what seems to be my standard estimate most days - depending on the task, I feel like I'm 50% to 200% more effective.  If you're mathing, that means I feel like I can increase my productivity by 3x if my robot colleague and I are in a groove.

As I pondered the question more, I began to realize what a crazy moment we're experiencing now.  Sure, different sectors of the economy grow at different speeds.  New areas are the most volatile but have the most growth capacity, while selling widgets at margin provides the dividends that Boomers dream of.  

But historically, in those hot sectors, investors would be on the lookout for a 10% increase in measured productivity.  A 3% increase across the entire economy would be an economic windfall.  Now though, the declared promise of AI is that a 10x increase in activity is the equivalent of me hand-collating 5-page memos and stapling them together as a teenager.  If you're not talking 100x increases or an approaching singularity, you're just manufacturing whips for the next version of horse they're constructing.

Seems reasonable.  

Until next time, my human and robot friends.