Noted. Thanks.

I first stopped reading after just the first sentence. The part "The rise of transformer-based architectures, such as..." is just click-bait.

But I re-opened and continued.

Then at the part where they offer a skewed definition of Intelligence, "First, a few definitions. Intelligence, as defined in this article, is the ability to compress data describing past events, in order to predict future outcomes...". This is not correct. Why not just use some agreed-upon definition? Like "The ability to acquire and apply knowledge and skills."

I'm just stopping there. Calling BS.

We have already seen too many articles where people imply AI will bring about loss jobs sending people into panic. Then on the other side we have this type of BS that causes fear and more panic.

Right!

In the case of a better AI model, I'd really like to have AI ask me the prompts instead of the current design where I have to ask the "right" questions to trigger the desired response.

For example, I post to AI, "I'm bored and want to read a book but not sure which one. Help me." Then it asked me a series of questions to get to some selection of available books I own or can acquire.

I can see for calculations and things with specific "correct" answers. But how will this handle the more "grey" area question where the answers are pretty subjective and as we've already seen dependant on the LLM data at the time?

Sweet! Thanks

Sorry, I meant to point to #11 on that sheet "ChatGPT scores IQ=147, 99.9th %ile." But the others are also interesting.

Is there a source for these?

Also wondering if ChatGPT or others have been subjected to a standard IQ test. Given that it was able to pass the Wharton and Medical exams.

I found this https://docs.google.com/spreadsheets/d/1O5KVQW1Hx5ZAkcg8AIRjbQLQzx2wVaLl0SqUu-ir9Fs/edit#gid=1264523637 but not the details source/owner of the sheet.

I kind of disagreed. If it were AI was used to make some stock pick predictions, I would then agree.

This is just that BuzzFeed announced there were going to use AI and humans decided to jump in to invest. Nothing more.

Are your expectations that all of society should just stop their day-to-day lives and use AI for everything??

"...ChatGPT has rolled out and the masses have been exposed to the power of AI"

The "masses" have not been exposed to the power of AI. First off, the ChatGPT system as it exists (version 3.5) is more or less a function prototype. Sure, there are many cool things that we can accomplish with it. But it is far from ready to take over things.

There are generally 5 stages for technology adoption: Innovators. Early Adopters. Early Majority. Late Majority. Laggards. I think we are somewhere between the "Innovators" and "Early Adopters" phases.

What is going to get us into the next phase will be the integration into common tools. This will be supported by Microsoft, Google, and others.

This somewhat follows the adoption of the Internet when it because publicly available in the early 1990s. You didn't see companies closing their storefronts to jump on the internet. It took the tinkerers to develop tools and standards. Companies eventually took the leap and published simple websites with static information mostly with address and phone numbers to their storefronts. It was a long time before things because more ubiquitous in our lives.