Discussion about this post

User's avatar
Ellis Reade's avatar

As one who has hit a few buckets of balls over the years, this post rings a lot of bells. And it brings to mind a line from David Foster Wallace in String Theory:

"…if both guys are good enough so that there are few unforced errors to break up the rally, a kind of fugue‑state opens up inside you where your concentration telescopes toward a still point and you lose awareness of your limbs … all you know then is the bright ball and the octangled butterfly outline of its trail across the billiard green of the court."

Some special moments with AI it feels like that - and for the rest, most shots go into the net or over the fence.

Expand full comment
Celeste Garcia's avatar

I have been confused since day one about the obsession with the perfect prompt. I don't remember when I began playing with LLMs but I downloaded GPT and input whatever was on my mind, and lo and behold, we had an exchange of ideas. A collab.

Every answer leads to more questions. I summarize in my own words to ensure I understand concepts and ask GPT to drill down on minutiae I don't understand. I push back when GPT says something trite (which is often) and call it out when it omits something important. Recently, while conversing with GPT about why Anthropic is pulling ahead of OpenAI in the Enterprise, GPT did not mention security. What? How can you talk about Enterprise Computing and not talk about Security? I called it out and GPT tried to cover for its blunder and then we engaged in depth on the topic.

None of my interactions with Gen AI required any specialized training, no directions in fine print necessary. Isn't that the point of AI? LLMs are built on decades of NLP (Natural Language Processing) research. We’re engaging with a system designed to interpret and respond to human language.

The Perfect Prompt rhetoric puts pressure on people who are already intimidated and feeling left behind by the barrage of AI everywhere. My advice: play with any of the generative AI models available. It's designed to be user-friendly. The technology itself is super complex, but we don't need to make using it into something esoteric and intimidating.

Expand full comment
14 more comments...

No posts