Artificial intelligence - limited

[Dave2002]

This article is interesting - https://getpocket.com/explore/item/w...the-real-world

I "liked" the comment here

Consider Tay, a Twitter chatbot released by Microsoft on March 23, 2016. Tay’s objective was to engage people, and it did. “What unfortunately Tay discovered,” Domingos said, “is that the best way to maximize engagement is to spew out racist insults.” It was snatched back offline less than a day later.

This suggests that the way to really get people involved is to insult them, to be really offensive.

I wonder if a former president had discovered the same or similar - not that some did.

Not a great idea!

[antongould]

It seems to be generally assumed that AI will, in the near future, save us …. but I feel we may have to hold our breath ….

Dealing, on behalf of a neighbour, with the dreaded EON I was trying to get detail on how her massively increased monthly instalment had been calculated .. in my first chat attempt I mentioned the price cap and an almost immediate response contained only a Janet and John explanation of the price cap. My second attempt majoring on the ignoring of my question stupidly mentioned standing charges …. again very swift response telling what standing charges cover …. but not answering my question …. how do I get away from AI and engage AHuman ….. ???

Coincidentally I set some friends a cricket question each morning and one says he uses an AI chat app with a reasonable degree of success so I have been having a play and find it very often wanders off on incorrect tangents and gives totalling incorrect information …. an example - I asked about the 1939 first-class cricket season and was very quickly informed there was no cricket that season because of the War ….. !!!!!

[smittims]

I think the general view in scientific circles is that we're still in the early stages, corrsponding to the motor car in the 1890s where it keeps breaking down and your neighbour shouts 'Get a horse!'

There was an hour-long discussion on the imminent benefits and hazards of AI on 'Woman's Hour' on Radio 4 this morning. It should be on BBC Sounds by now. The news was that it's about to make huge strides in what it can and will do.

I'm not a fan, as I'm still not convinced that the telephone and the aeroplane have brought more good than harm . But it seems AI is coming, whether we like it or not.

[oddoneout]

It amused me to read an article recently reporting that reading the manual could help AI learn more quickly. This https://opendatascience.com/new-tech...ional-methods/ isn't the article but sums up the findings.
RTFM as they say in certain circles

[Dave2002]

I recently had occasion to call a help line regarding a problem. The “person” at the far end seemed so slick with questions and responding to answers that I began to suspect I was dealing with a robot voice system. Eventually I asked the question - “are you a robot?” [ and I tried, I hope politely, to explain that the interaction up to then had been so quick that I suspected some automation] after which we had quite a pleasant conversation. Checking out the Turing test in real time is something we may have to try out more often in the future.

[Anastasius]

....
Dealing, on behalf of a neighbour, with the dreaded EON I was trying to get detail on how her massively increased monthly instalment had been calculated .. in my first chat attempt I mentioned the price cap and an almost immediate response contained only a Janet and John explanation of the price cap. My second attempt majoring on the ignoring of my question stupidly mentioned standing charges …. again very swift response telling what standing charges cover …. but not answering my question …. how do I get away from AI and engage AHuman ….. ???

....

Email the CEO ? ceo@eonnext.com

[Anastasius]

I posted a negative review on Trustpilot regarding Aviva's rejection to provide me with life insurance even though the current policy with them had just ended and nothing detrimental had happened regarding my health. They picked up the review and replied with the usual vapid boilerplate reply..."so sorry...blah...blah..blah" followed by "Please let us know your policy number ad we will look into it for you".

I enjoyed adding a postscript to my original review in response.

[french frank]

Email the CEO ? ceo@eonnext.com

I did that once with BT, following advice located in a D Telegraph article. No response from the CEO, but I did get a personal phone call from a chap with a name (George, I think), and it was sorted.

[Dave2002]

I did that once with BT, following advice located in a D Telegraph article. No response from the CEO, but I did get a personal phone call from a chap with a name (George, I think), and it was sorted.

Ah yes! What company is Elon Musk the CEO of? Maybe Twitter, perhaps Tesla, and probably a few others. I’m still awaiting the courtesy of a reply to a letter sent directly to him in January.

[Old Grumpy]

Ah yes! What company is Elon Musk the CEO of? Maybe Twitter, perhaps Tesla, and probably a few others. I’m still awaiting the courtesy of a reply to a letter sent directly to him in January.

Ah, but did you start the letter with 'Dear A**hole'?

[Anastasius]

I did that once with BT, following advice located in a D Telegraph article. No response from the CEO, but I did get a personal phone call from a chap with a name (George, I think), and it was sorted.

That's the whole point. All the Exec offices have teams who respond to these emails. They kick butt. Problem sorted.

But they need to be aware of the issue.....hence the CEO email

[Anastasius]

Ah, but did you start the letter with 'Dear A**hole'?

Yebbut...SpaceX ...you have to admit that watching those launchers land back on earth, from an engineering PoV was pretty cool

[Dave2002]

Yebbut...SpaceX ...you have to admit that watching those launchers land back on earth, from an engineering PoV was pretty cool

Some of the SpaceX stuff is/was amazing, thought not all worked, some landed, but then blew up. The motors in Tesla cars are very good. The customer service by the Tesla company is abysmal. The cars are not particularly well adapted to driving on British roads, or in British weather, and have more faults than I can count on my fingers - of one hand at least.

[antongould]

It seems to be generally assumed that AI will, in the near future, save us …. but I feel we may have to hold our breath ….

Dealing, on behalf of a neighbour, with the dreaded EON I was trying to get detail on how her massively increased monthly instalment had been calculated .. in my first chat attempt I mentioned the price cap and an almost immediate response contained only a Janet and John explanation of the price cap. My second attempt majoring on the ignoring of my question stupidly mentioned standing charges …. again very swift response telling what standing charges cover …. but not answering my question …. how do I get away from AI and engage AHuman ….. ???

Coincidentally I set some friends a cricket question each morning and one says he uses an AI chat app with a reasonable degree of success so I have been having a play and find it very often wanders off on incorrect tangents and gives totalling incorrect information …. an example - I asked about the 1939 first-class cricket season and was very quickly informed there was no cricket that season because of the War ….. !!!!!

Sadly the interface continues - I asked for the kWh rates and consumptions used in the instalment calculation and seizing on the mention of rates AI trotted out the rates we can see on my friend’s bill - nothing else in what it seemingly considered a fine response ……. I will soldier on perhaps ending at the CEO’s office ……

[Dave2002]

To heck with AI - currently have a problem up in Scotland with CalMac. The company seems completely incapable of dealing with problems due to the failure of its ships, using AI, human intelligence, or otherwise