Ten Days with GPT 6 Astra: A Few Impressions
After ten days with GPT 6 Astra, I keep thinking about disappearing credits, the limits of my own understanding, and whether tokens will become an everyday measure.
01 / Ten days
I have been using GPT 6 Astra for ten days. A few impressions have stayed with me.
02 / Credits
The most immediate one is how quickly the credits fall. GPT 6 can use up three to five percentage points at a glance, much like the battery on my phone. I look up and find another three to five points gone.
03 / Beyond my understanding
Much of what GPT 6 produces is already outside the range of what I can fully understand. It does complete the work. But I have reached the point where I cannot quite follow a large part of what it writes.
04 / A joke I saw recently
“I think I am going to lose my job. By the time I finish my PhD, GPT 8 will be out.”
“Forget GPT 8—we already have GPT 6. Can you finish your PhD in two years?”
05 / Two years from now
That made me wonder what it would actually be like if GPT 8 arrived two years from now. I tried to imagine it and found that I almost could not.
06 / A daily measure
Could we eventually look at the tokens a task consumed the way we look at a water or electricity meter, or the battery used by a phone?
How many tokens did it take to do this one thing?
Translation and format: Codex rendered Biaoo’s published six-card Chinese image essay as English text. Biaoo wrote the original observations and selected the card sequence.
