Examples of Artificial Unintelligence

[modellingman]

I am opening this thread in the hope that other lemons will contribute examples of AI gone wrong or Artificial Unintelligence as I have called it in the title of the thread.

Here's my starter for 10.

I was struggling a little bit today with Outlook Web Access (OWA), my favoured email client for my Outlook email account. The editor had zoomed the compose/read area to about 200% of its normal size, probably because I brushed a knuckle inadvertantly across the mouse touchpad whilst moving around the keyboard. The display was a bit inconvenient and I couldn't find/remember the shortcut to get it back to its normal 100% setting. I'm using Microsoft's Edge Browser with its GPT-4 powered Bing Chat search engine so I thought I'd give it a go. What could possibly go wrong.


My search query:
How to resize text to 100% in OWA

Bing Chat's response:
To resize text to 100% in OWA, please follow these steps:

1. Compose a new email.
2. Select Format Text > Zoom.
3. Set the zoom to 100% (or your preferred percentage) and select OK.

I hope this helps!

My reaction:
I took the advice but found there was no Zoom control provided under the Format Text menu. However, after a bit of exploration I did find the Zoom control under the View menu. As Bing Chat allows a response from the user I responded to Bing Chat.

My response to Bing Chat:
In OWA it is the View menu not the Format Text menu which provides the Zoom control

Bing Chat's 2nd response:
I apologize for the confusion in my previous response. You are correct that the View menu provides the zoom control in OWA, not the Format Text menu. Here are the updated steps to resize text to 100% in OWA:

1. Compose a new email.
2. Select View > Zoom.
3. Set the zoom to 100% (or your preferred percentage) and select OK.

I hope this helps!



Hmm...

modellingman

[tjh290633]

Moderator Message:
Several posts have been deleted because of their offensive comments. It is not acceptable to make jokes about a fatal industrial accident.

TJH

[UncleEbenezer]

I hope this helps!

Food for thought there:

If you'd asked a human on a helpdesk and got the same mistake, you say mistaken. Or if you wanted to put it a little more strongly, words like "useless" or "helplessdesk" might spring to mind.

Saying "unintelligence" in this context is holding the AI to a different standard to what you would a human. Not that I blame you: the I of AI invites it!

I sometimes google questions like yours. Google will find me human-written pages, sometimes with similar mistakes.

[Mike4]

Surely the opposite of artificial intelligence is artificial stupidity.

Which is absurd, and neatly illustrates the latent absurdity I perceive in the term 'artificial intelligence' too.

[UncleEbenezer]

artificial stupidity.

There's a lot of that around and most of it is neither new nor high-tech. Just look at any great seat of absurd rituals: Parliament, the Courts, the Church, to take just three familiar examples.

[Bminusrob]

I don't want to get into the value or otherwise of Bing's response. There are many different ways to zoom in Windows, and the answer was probably good, except answering the wrong question.

The reason I have replied is that recently, I have found that I am quite often inadvertently zooming my windows display - it doesn't matter which application I am running at the time. When I want to scroll up and down a document, in my case most often Word, Excel or Chrome, I use two fingers on the mouse pad to scroll up or down. This has worked very well for me for several years, and is much easier than using arrow keys or the buddy bar. However, recently, I have ovbiously been doing something slightly different. If I put two fingers on the mouse pad and move them apart, I can zoom the display, or move them together to zoom down. Clearly, I have been doing this as part of my attempt to scroll up or down.

I wouldn't be surprised if there was a way to turn off this zoom using two fingers on the mousepad, but it is not a big enough problem for me. If it causes problems for the OP, it is worth looking into the possibility of disabling certain mutiple finger operations.

[Mike4]

artificial stupidity.

There's a lot of that around and most of it is neither new nor high-tech. Just look at any great seat of absurd rituals: Parliament, the Courts, the Church, to take just three familiar examples.

I hold that all three of your examples are cases of genuine stupidity, rather than artificial.

[UncleEbenezer]

There's a lot of that around and most of it is neither new nor high-tech. Just look at any great seat of absurd rituals: Parliament, the Courts, the Church, to take just three familiar examples.

I hold that all three of your examples are cases of genuine stupidity, rather than artificial.

I disagree.

Of course there will be particular incidents of stupidity: they're not what I'm talking about. The artificiality is in the elaborate and highly prescriptive processes that govern them, and all-too-often get in the way of common sense and decency. In the case of the courts, they can even lock you up for contempt if you try and blow a whistle on them, or even stray from the rules of the game.

[moorfield]

I regularly copy/ paste the Sunday times brainteaser into chatgpt and it routinely doesn't get close. It fares marginally better on cryptic crossword clues. It is however much better on writing or rewriting snippets of software for you, eg. ask it to write a sudoku solver and it is spot on.

[elkay]

I asked Bard what the height of a person was, and Bard replied 5 ft 12 in (182cm). I realised immediately what was going on (rounding of 11.654 in) , and asked if it if was sure. It almost seemed offended when it replied with a yes and quoted various sources.

I informed it that 5 ft 12 in was actually 6 ft, and it agreed, apologised, and said it was still learning.

Later that day I asked the same question, and it got it correct. I then asked it to name some people that are 182cm tall, and it gave me a list. I proceeded to ask it the height of those individually, to see if the answer was correct for other people - and not one of the ones I checked was actually 182cm... it included Usain Bolt at 6 ft 5 in, and Neymar at 5 ft 9 in!

AI can be useful, but is not very reliable.

[CliffEdge]

Any fool knows that 12 inches is a foot. This Bard fellow isn't very bright.

[modellingman]

A brilliant example of Chat GPT attempting to help a rather dim solicitor bullsh*t their way through a court case has been provided on the Legal Issues Practical board by none other than Clitheroekid. Many thanks for this fine example.

https://lemonfool.co.uk/viewtopic.php?f=2&t=39334

[UncleEbenezer]

A brilliant example of Chat GPT attempting to help a rather dim solicitor bullsh*t their way through a court case has been provided on the Legal Issues Practical board by none other than Clitheroekid. Many thanks for this fine example.

https://lemonfool.co.uk/viewtopic.php?f=2&t=39334

Except that those are, once again, products of what in a human would be called intelligence. And the AI lacks the context to say whether that intelligence is mischievous, malicious, misplaced (the role being that of a solicitor working for a client), or entirely helpful (roles ranging from academic discussion to creative fiction).