The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
ChatGPT-4 Sharply Outperforms ChatGPT-3. But Humans-4 Don't Sharply Outperform Humans-3
AI technology keeps improving. NI, on the other hand, is staying about the same.
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
Yeah, I’m actually glad I’ll be retiring in a few years; While I comfortably beat GPT-4’s SAT scores, I kind of suspect that GPT-5 will beat mine, and not just because my brain is aging. Technological obsolescence is coming for those of us who work with our brains, not brawn, finally.
I’d honestly stopped expecting the singularity to happen in my lifetime. Color me surprised.
When the programs learn how to program themselves to get smarter and better...that's when you need to worry.
It can already program, just not so well. The AI progression from "Can, but not well" to "World beating" is pretty short, though.
The thing is, they seem to have hit on an approach to AI which rather that relying on clever programming and a deep understanding of intelligence, just depends on throwing a huge amount of resources at the problem, and the intelligence seems to scale with the resources devoted, rather than running into diminishing returns.
That means maybe you can't mass produce superhuman AI's, but building one Colossus might be within reach.
Neural networks and genetic algorithms already do.
Put information in computer. Computer puts information back out. Amazing.
Are humans-4 a downgrade?
"Please never forget what artists were once capable of."
https://twitter.com/Culture_Crit/status/1635740138243325971
The hardware is still fine, but they've been rolling out some degraded software recently.
Humans-4 are actually better than Humans-3 at IQ tests. It’s called the Flynn effect (https://en.wikipedia.org/wiki/Flynn_effect). No one understands why.
Of course AI is progressing at a much faster pace than NI is.