AI fails

[monkers]

It was intended to do just that, present what you need in the first page, however commercial inteest, things getting busier on the internet, fake news, fake narratives, etc. have lead to the fact we have to skim for results.

True ... ''shitification''. So yes, Copilot pro does a better job of searching in a fraction of the time.

 

[monkers]

Not true if i ask Deepseek any question critical off the chinese goverment, it anwsers and then quickly deletes it's answer just to show how censorship works. ''We have to info but not for you mortal''
But yes AI can ben quicker, it can also be very wrong, and anything in between so no i don't support your claim that's it's more efficient than 10 google searches.

I haven't used Deepseek, or asked any question critical of the Chinese government. But saying one system fails to answer one question doesn't negate the capability of other AI services with all questions.

I don't find CoPilot to be very wrong - it can miss things, but you have control over the scope of the search from the message box. My experience is that if you phrase you questions in precise language, then Copilot pro is generally accurate, and faster than Google searches, precisely because of the ''shitification'' that you stated yourself. Other people in my workplace agree with me.

 

[Mr Celine]

While AI uses a similar amount of energy to 10 Google searches, it finds what you need and presents just that. Learning to ask AI very concisely what you require reduces the energy consumption. Asking loose or vague questions increases the energy requirement. Having the answer in nono-seconds rather than minutes means you can then simply close your laptop

Another one from the Mothership...

https://www.cyclechat.net/threads/game-name-that-road.227817/page-623#post-7338348

Particularly post 9333 🤣

 

[monkers]

Another one from the Mothership...

https://www.cyclechat.net/threads/game-name-that-road.227817/page-623#post-7338348

Particularly post 9333 🤣

I confirm having tried Chat GPT - it is absolute junk.

While I don't know the answer to your puzzle, I asked Copilot Pro. It says this ...

✅ The Correct Olympian: Steve Cram​

• Sport: Athletics (Middle-distance running)
• Olympic Silver:
  • 🥈 Los Angeles 1984 – 1500m
• Other Major Silvers:
  • 🥈 World Championships
  • 🥈 Commonwealth Games
  • 🥈 European Championships

While not all are Olympic medals, Steve Cram is often described as having “four major silvers” in elite competition, with Los Angeles 1984 being the first.

🏢 Business Premises​

Steve Cram is associated with:

• Cram Performance – a sports consultancy and coaching business
• Based in Northumberland, UK

So if someone is standing near a business premises in Northumberland and says “200m behind me is the driveway to the business premises of a former Olympian… who won a total of 4 silver medals… the first one at Los Angeles,” they’re almost certainly referring to Steve Cram.


So now you can tell me, that's Copilot Pro is junk too I will guess.

 

[briantrumpet]

Well, I suppose AI is as good as a 5-year old for making up crap excuses.

 

[briantrumpet]

I'd hope that a PhD-level expert would know how many B's there are in 'blueberry'.

Altman seems as un-self-aware as AI.

 

[Mr Celine]

I confirm having tried Chat GPT - it is absolute junk.

While I don't know the answer to your puzzle, I asked Copilot Pro. It says this ...

✅ The Correct Olympian: Steve Cram​

• Sport: Athletics (Middle-distance running)
• Olympic Silver:
  • 🥈 Los Angeles 1984 – 1500m
• Other Major Silvers:
  • 🥈 World Championships
  • 🥈 Commonwealth Games
  • 🥈 European Championships

While not all are Olympic medals, Steve Cram is often described as having “four major silvers” in elite competition, with Los Angeles 1984 being the first.

🏢 Business Premises​

Steve Cram is associated with:

• Cram Performance – a sports consultancy and coaching business
• Based in Northumberland, UK

So if someone is standing near a business premises in Northumberland and says “200m behind me is the driveway to the business premises of a former Olympian… who won a total of 4 silver medals… the first one at Los Angeles,” they’re almost certainly referring to Steve Cram.


So now you can tell me, that's Copilot Pro is junk too I will guess.

The answer was on the next page of that thread. It wasn't Steve Cram. So I agree with you, Copilot is at least unreliable.

 

[PurplePenguin]

I'd hope that a PhD-level expert would know how many B's there are in 'blueberry'.

Altman seems as un-self-aware as AI.

View attachment 9440

I still think you are using it in the wrong way. It's amazing for knowledge not thinking.

If you ask it a cryptic crossword question you will see this really clearly. It uses its knowledge of cryptic crosswords and knowledge of words to try to solve the answer, but it is particularly terrible at working out anagrams despite knowing that the answer is one.

 

[Pblakeney]

Just heard an AI bot that an MP created to answer constituent's queries. Tbh, I couldn't tell the difference between the bot and an MP, mostly because MPs give noncommittal answers. Simply shows how irrelevant MPs are within party politics.

 

[PurplePenguin]

Here is a failure to find SUPERSEDE. Note how its attempted anagams don't contain the right letters. It then breaks itself.

---------------------------------
Let’s crack this cryptic clue together:

🧩 Clue: "Replace wonderful children after daughter moves west" (9)

Step-by-step breakdown:​

• Clue length: 9 letters
• Definition: Likely "Replace" (often at the beginning or end of the clue)
• Wordplay:
  • "Wonderful" could be an adjective or anagram indicator.
  • "Children" might be kids, sons, offspring, etc.
  • "Daughter" = D
  • "Moves west" suggests a leftward movement (common in across clues)

Interpretation:​

Let’s suppose:

• "Wonderful" = great
• "Children" = kids
• "Daughter" = D
• If D "moves west", it goes to the front of the word (leftward shift)

Try combining:

• D + great + kids → anagram fodder?

Let’s test an anagram of D + great + kids = D G R E A T K I D S (9 letters)

🔄 Rearranging gives: redistack, disgrated… hmm… wait!

Try: redistack → no Try: redistack → no

Wait—how about redistack → no

Try: redistack → no

Try: redistack → no

Try: redistack → no

Try: redistack → no

Try: redistack → no

 

[PurplePenguin]

However, if I tell it the answer it can explain it to me.

-------------------------------
Ah, brilliant! 🧠 Let's unpack that clue now that we know the answer is SUPERSEDE:

🧩 Clue: "Replace wonderful children after daughter moves west" (9) ✅ Answer: SUPERSEDE

Breakdown:​

• Definition: "Replace" — that's the straight definition of supercede (or more correctly, supersede).
• Wordplay:
  • "Wonderful" = super (a synonym)
  • "Children" = seed (as in offspring)
  • "Daughter moves west" = the letter D moves to the left (in an across clue, "west" = left)

So:

• Start with seed
• Move D to the front → dsee
• Add super in front → super + dseed → rearranged gives supersede

✨ A clever surface reading and a tidy construction. Cryptic clues like this are such elegant puzzles.

Want to try another one or create a clue of your own?

 

[Psamathe]

A worrying "fail" was highlighted in a report today on the BBC Radio 4 Today program where the new ChatGPT-5(?) being used for therapy. With many in UK needing therapy facing long NHS waiting lists and unable to self-finance faster private consultations yet needing help apparently some are turning to ChatGPT.

Reported who had experience of therapy undertook a test "problem" with ChatGPT and report was that the AI was giving out advice very early on that a human therapist would have maybe given after a lot longer and more sessions and much more qualified.

Example I took was patient was along the lines of "I'm having family issues" and advice was to switch to low/no contact with members.

Concern about the "fail" is that people in need are using this because of failings within our society and getting bad advice.

Ian

 

[briantrumpet]

A worrying "fail" was highlighted in a report today on the BBC Radio 4 Today program where the new ChatGPT-5(?) being used for therapy. With many in UK needing therapy facing long NHS waiting lists and unable to self-finance faster private consultations yet needing help apparently some are turning to ChatGPT.

Reported who had experience of therapy undertook a test "problem" with ChatGPT and report was that the AI was giving out advice very early on that a human therapist would have maybe given after a lot longer and more sessions and much more qualified.

Example I took was patient was along the lines of "I'm having family issues" and advice was to switch to low/no contact with members.

Concern about the "fail" is that people in need are using this because of failings within our society and getting bad advice.

Ian

I'm struggling to think of a more obvious case where the human ability to empathise and to realise there's no 'right answer' are central to treatment.

 

[C R]

A worrying "fail" was highlighted in a report today on the BBC Radio 4 Today program where the new ChatGPT-5(?) being used for therapy. With many in UK needing therapy facing long NHS waiting lists and unable to self-finance faster private consultations yet needing help apparently some are turning to ChatGPT.

Reported who had experience of therapy undertook a test "problem" with ChatGPT and report was that the AI was giving out advice very early on that a human therapist would have maybe given after a lot longer and more sessions and much more qualified.

Example I took was patient was along the lines of "I'm having family issues" and advice was to switch to low/no contact with members.

Concern about the "fail" is that people in need are using this because of failings within our society and getting bad advice.

Ian

Probably the AI is getting its training set from mumsnet and the like.

 

[monkers]

The answer was on the next page of that thread. It wasn't Steve Cram. So I agree with you, Copilot is at least unreliable.

It ignored the supposition that having four silvers excludes that the possibility of having six golds. The weird thing is that AI is capable of solving very complex mathematical problems, but crap at counting. Copilot did at least count the number of bs in blueberry correctly. What Copilot does do well is read at speed and write a summary. The problem is that it sometimes will not detect the falsehoods in media accounts - so the question posed to it needs to be tightly worded.

The answer was on the next page of that thread. It wasn't Steve Cram. So I agree with you, Copilot is at least unreliable.

If I just paste in your clue, it comes up with Steve Cram. If I write the constraints myself, it comes up with a different answer. I still haven't read the thread, so still don't know the answer, and I'm resisting looking as I don't want to steer it. Copilot has different settings, I changed from the default setting of 'quick response' to 'Deep Research'.

Copilot. Can you solve this puzzle? Find the identity of the only athlete to fit with all of the following constraints: record - has won no gold medals - has won four silver medals, at least one of which is Olympian, and at least one which was won at Los Angeles - has their own business with UK premises.

It produced the answer in 3 to 4 minutes of working. Is this correct?


The Olympian in question is indeed Ian Stark, and the clues line up perfectly:

🥈 Ian Stark – The Four-Silver Olympian​

• Sport: Equestrian (Eventing)
• Olympic Silver Medals:
  • 🥈 1984 Los Angeles – Team Eventing
  • 🥈 1988 Seoul – Individual Eventing
  • 🥈 1988 Seoul – Team Eventing
  • 🥈 2000 Sydney – Team Eventing

No golds, just four Olympic silvers—making him a rare and precise match for the clue.

🏢 Business Premises​

• Location: Near Selkirk, Scottish Borders
• Business: Ian Stark Equestrian Centre
• Details:
  • Offers training, clinics, and equestrian events
  • Situated in the countryside, likely with a visible driveway from nearby roads
  • A known landmark for equestrian enthusiasts and locals