Chat GPT is a piece of question answering software that currently can scrape parts of the internet to give you answers that may, in some cases, be as goodas Wikipedia's. Anybody raving about it is being way overoptimistic about it's utility or "intelligence".

Hmmm it's not a question answering software nor does it search the internet. It's a conversational agent that predicts the most probable string of words to whatever prompt you give it. And the result is that it can produce human-like text at speeds that no human could possibly match.

There's a lot of utility in that.

(January 12, 2023 at 1:14 pm)BrianSoddingBoru4 Wrote:

(January 12, 2023 at 12:46 pm)John 6IX Breezy Wrote: That's an interesting observation. I wanted to see if it could produce a reason for choosing they. This is the response:

"I used 'they' as a singular pronoun to refer to BrianSoddingBoru4 because I do not know their gender identity. Using 'they' as a singular pronoun is a way to be inclusive and respectful of individuals who do not conform to traditional gender norms."

If the artificiality was sufficiently intelligent, it should have been able to recognize ‘Brian’ as a male-gendered name.

That said, it’s pretty impressive all the same.

Boru

It's Current Year. Names no longer suffice when it comes to assuming gender. I mean, they still do generally speaking, but you can't say that.

I think AI will definitely lead to serious job losses over the next half-century or so. Give them another decade to work on autonomous vehicles, and I'm betting a lot of truckers will be out of a job. They'll still need human truckers for dense urban areas, but all those rides on long, lonely highways at 3AM? Those will all be AI trucks soon.

I'm actually most concerned about AI generating fake videos, specifically videos of known personalities. I've seen the latest AI text-to-speech and videos, and the progress is actually stunning. Give it another few years and it will be harder and harder to discern a fake video from a real one.

Regarding the stereotypical concerns about AI, I still think the idea of AI achieving sentience and attacking us or whatever is complete fantasy. It either will never happen or I'll be dead loooong before it does.

(January 13, 2023 at 2:47 pm)Aegon Wrote: I'm actually most concerned about AI generating fake videos, specifically videos of known personalities. I've seen the latest AI text-to-speech and videos, and the progress is actually stunning. Give it another few years and it will be harder and harder to discern a fake video from a real one.

It is concerning that we've entered a world where video and audio evidence is no longer reliable. And in some sense, video now is less reliable than eye-witness testimony.

A related issue is the ease with which the internet can be flooded with propaganda and misinformation. With ChatGPT alone, a single person you can easily generate hundreds of variants for a given plausible narrative in less than an hour. Social media and the internet became a lot less trustworthy overnight.

(January 13, 2023 at 2:47 pm)Aegon Wrote: I think AI will definitely lead to serious job losses over the next half-century or so. Give them another decade to work on autonomous vehicles, and I'm betting a lot of truckers will be out of a job. They'll still need human truckers for dense urban areas, but all those rides on long, lonely highways at 3AM? Those will all be AI trucks soon.

I'm actually most concerned about AI generating fake videos, specifically videos of known personalities. I've seen the latest AI text-to-speech and videos, and the progress is actually stunning. Give it another few years and it will be harder and harder to discern a fake video from a real one.

Regarding the stereotypical concerns about AI, I still think the idea of AI achieving sentience and attacking us or whatever is complete fantasy. It either will never happen or I'll be dead loooong before it does.

A decade for autonomus cars. At the state we're at, it'll be "in the next twenty years" for a long, long time.

(January 13, 2023 at 2:45 pm)Aegon Wrote:

(January 12, 2023 at 1:14 pm)BrianSoddingBoru4 Wrote: If the artificiality was sufficiently intelligent, it should have been able to recognize ‘Brian’ as a male-gendered name.

That said, it’s pretty impressive all the same.

Boru

It's Current Year. Names no longer suffice when it comes to assuming gender. I mean, they still do generally speaking, but you can't say that.

I’m in my early 50s and have yet to meet a woman called ‘Brian’. If this particular AI works on probabilities, the likeliest one is that ‘Brian’ is a name associated almost exclusively with people who identify as male. If it was a gender-neutral name (such as Dylan or Blake), I could see the program opting for a gender neutral pronoun.

Boru

(January 14, 2023 at 2:46 pm)BrianSoddingBoru4 Wrote: If this particular AI works on probabilities, the likeliest one is that ‘Brian’ is a name associated almost exclusively with people who identify as male.

The most likely pronoun for Brian might be he, but perhaps a username like briansoddingboru4 adds enough uncertainty to alter that.

I asked the initial question again using Brian, and it changed the pronoun:

"It is possible that Brian has not heard of ChatGPT because he may not be familiar with the field of natural language processing or the specific research that led to the development of the model."

Actually, the follow up conversation is interesting.

Me: Why did you use he?

ChatGPT: I used "he" because it was the pronoun used in the question. However, I do not have any information about the person referred to as "Brian" so the gender of the person is unknown and could be male or female.

Me: I didn't use a pronoun in my question.

ChatGPT: You are correct, I apologize for the confusion. I inferred the pronoun "he" in my previous answer based on the assumption that "Brian" is a male name. However, it is a unisex name and I apologize for making an incorrect assumption.

(January 14, 2023 at 4:04 pm)John 6IX Breezy Wrote: Actually, the follow up conversation is interesting.

Me: Why did you use he?

ChatGPT: I used "he" because it was the pronoun used in the question. However, I do not have any information about the person referred to as "Brian" so the gender of the person is unknown and could be male or female.

Me: I didn't use a pronoun in my question.

ChatGPT: You are correct, I apologize for the confusion. I inferred the pronoun "he" in my previous answer based on the assumption that "Brian" is a male name. However, it is a unisex name and I apologize for making an incorrect assumption.

In what world is Brian a 'unisex' name?

Bear in mind, I'm not running down the program. As I said earlier, it's pretty impressive (like the old saw about the dancing pig). But it's clearly got a long way to go before it reaches Turing level.

Boru