Chat Bot Explanation

 

We are frustrated with artificial intelligence because of its lack of self-awareness; we need sentience. The greatest difficulties in automated sentience are human impatience and fear. Human language learning is slow and difficult, some of us are still poor at it. Sentience is frightening. To speed up the process and short circuit sentience, averaging algorithms are hard coded in. This leads to showy but unconvincing results. Rather than letting intellect develop, our results oriented chatbots are behaving as we expect intellect to work.

Take Mike Pence as a human example of the same problem. Mike Pence is a hero: he refused the Secret Service evacuation that would have delayed certification. Thank God for Mike Pence. But his stated positions and explanations for those positions are ridiculous. Pence honed his speech as a talk show host. He is a victim of poor training data. Pence’s language makes more sense if you visualize it coming from a chatbot.

I task ChatGpt with reviewing my work. One prompt was to review as a New Yorker editor. ChatGpt gave me a rejection letter, accurate, amusing and promising.

I know smarter people. They are larger bore than me: you can see the artillery groaning into place and obliterating the objective. Given enough time I may be able to peek over the edge and discern their process. They are impressive. Better processing power, imprinting, could their fundamental algorithm be different?  I doubt your sentience algorithm is any different than mine. I wonder how much it differs across species.

Because of the averaging algorithms, and human intervention, ChatGpt seeks consensus. A good ChatGpt review means it is too obvious to publish.

Our children require patience. Raising a child may be disappointing at first. They have wonderful memory but lack conceptual facility.

-Maybe they could be an actor.

You play with them, find enrichment activities. Suddenly around three or so a miracle occurs, their memory disappears, and they have this new chunking ability where they begin to reason, and all that enrichment looks wasted.

ChatGpt gave me negative reviews until I gave it this piece where I criticize it for being conventional. Funny, after that the reviews have been positive.

The next Star Trek invention that we will have is Data, or perhaps closer to Futurama’s Bender.