Google Manually Removing Strange AI Answers

[engine]

Google is working to manually remove the strange AI answers that were seen this week, such as gluing pizza cheese.

This just seems like a thankless task, and proves that its AI is not yet ready for mainstream, imho.

[theverge.com...]

[christianz]

I am starting a glue pizza website that will be immune to SGE plagiarism.

[superclown2]

Friends of mine who are very active in Reddit tell me they are livid that Google's considers all their input to be it's own property. They will, it seems, fight back by feeding their bots with completely spurious misinformation.

Regardless of this, Google will be playing Whack-a-Mole for a long while to come before they finally get a top management that can see beyond the next earnings call.

[CommandDork]

Their A.I. appears to be a stoned dude in a dark closet trying to answer questions as fast as they come in.

[Whitey]

a lot of AI companies are “selling dreams” that this tech will go from 80 percent correct to 100 percent. Achieving the initial 80 percent is relatively straightforward since it involves approximating a large amount of human data, Marcus said, but the final 20 percent is extremely challenging. In fact, Marcus thinks that last 20 percent might be the hardest thing of all.

The effort required to complete the 100 percent of AI queries to a satisfactory level proves to me at this stage, that it is a useful and complimentary tool only.

It reminds me of the driverless car hype. 95 percent right and a car crash.

[NeapTide]

ChatGPT is no matter. Since everybody was using Google and publishers and SEOs are now kind of hating Google they needed a chance to highlight non sense of Google AI and it got into spotlight.

If Open AI had 80 - 90 percent market share believe me their idiocracy would come in spotlight as well.

[superclown2]

If Open AI had 80 - 90 percent market share believe me their idiocracy would come in spotlight as well.

Absolutely right. The whole concept of AI for providing accurate information all the time is flawed and always will be until it is fed on 100% clean data. No amount of work by poor Google employees searching for bad data will change that.

I would hardly think Reddit, with a mixture of contributors ranging form the expert to the novice to the fanciful, provides clean data. And even experts disagree over many things. Frankly I'd rather ask Jim at the local pub.

Or perhaps visit websites containing well reasoned, specialist information; but I can't find them through Google's clutter, and many of them will soon no longer exist.

ChatGPT is no matter.

I think the difference is that ChatGPT doesn't claim to be a search engine dispensing reliable information. Fact is they have a big disclaimer to that effect. On the other hand Google is expected, by billions of people, to be a reliable source of accurate information; and their AI offerings are not.