Hello,

Read this:
https://www.malwarebytes.com/blog/n...d_v2_168867428049&utm_content=google_train_ai

Bertus

 

Everything you've ever posted as already been used to train AI. Google is just making it formal.
https://www.diffbot.com/

 

Considering my œuvre, that is going to be one weird AI.

 

Wow, all my mistaken posts will be immortalized!

 

Only Google?

 

Can't wait until AI incorporates all the on-line conspiracy theories, fake stories, pseudo science, and "alternate facts" into its memory to spew back at us.
Should make for some interesting (if not downright scarry) reading.

Methinks the light at the end of the tunnel is the AI train coming straight at us.

 

Can't wait until AI incorporates all the on-line conspiracy theories, fake stories, pseudo science, and "alternate facts" into its memory to spew back at us.

No need to wait -- it's already in there.

 

Since Google's engineers (and other groups working on this) are well aware of the unreliability of so much that is on the Internet I think it is safe to assume that part of the project is trying to find a way to distinguish the reliable from unreliable.

It is an interesting problem, and the first approaches I can think of necessitate identifying examples of each sort for the training dataset. This means that it will start with some "appointed authority". A very human problem indeed.

I am interested, if this is being done (well) to see how the model deals with marginal cases where it is not conspiracy theory nonsense but instead is just new science or new technology that defies the ordinary expectations and isn't being treated seriously.

How would such a model have treated Einstein's Annus Mirabils papers? I would assume just like his colleagues who ranged from interested, to skeptical, to dismissive, to outraged. But in this case, if people think (as they seem to be doing with ChatGPT) that it "understands" all the complicated technical stuff they don't.

And with the presence of the Internet allowing utterly uninformed people to assert their opinions on topics they've no business having an opinion on, would Einstein have been "cancelled" by the mob on Twitter for his "pseudoscience" by the righteously outraged horde of wannabe science sophisticates?

Of course, the first part of my speculation is based on a very naïve and trivial approach to the problem so maybe they will have a more clever way of producing the differential. If so, maybe it will be effectively "more expert than the experts" at least in identifying the patterns that tend to indicate a breakthrough hidden to the biased minds of the (genuine) experts who have many reasons to reject innovative approaches to their fields.

If the latter, it could be very interesting and possibly game-changing.