Try this one: A right-angled triangle has base length 10 and height 6
measured from the hypotenuse to the opposite apex (the base is the
hypotenuse). What's the area of the triangle?

ChatGPT says:

For any triangle (right-angled or not), the area is:

Area = 1/2 base × height

So the answer is 10 × 6 / 2 = 30.

Wrong!  It's a nasty catch. It goes awry at the very first statement: the
expression given is true for some triangles but not all, and in particular
not this one.  In fact it's not even a triangle so the area is undefined:
that's the correct answer.  It's impossible for the height to be more than
5 (draw the circumscribing circle).  It fell into the trap of unthinking
application of a standard formula fetched from the web without determining
first whether it applies to the situation (and I imagine most humans will
say 30 too without thinking!).

This was a question in a job interview at Google.

-- Ian


On Sun, 25 Jan 2026, 15:31 Goldman, Adrian, <
[email protected]> wrote:

> I think it really depends on use case. It’s extremely good at removing
> english language mistakes, for instance. But at heart it’s a statistical
> model, right?  What word is likely to follow ‘what’ in a sentence about
> ChatGPT, for instance. Like my last.  The likelihood of painterly is very
> close to zero, but is, word, token have all got high probabilities. So for
> things it should be good at - it really is good.
>
> That doesn’t include facts: the intersection of words with the world.
>
> chatgpt5 does facts slightly better than chatgpt2. I remember asking it
> for sonnets and 2 hadn’t got a concept of sonnet - but 3+ do. You’ll get 14
> lines in one of the classic sonnet rhyme patterns in iambic pentameter.
>
> But don’t worry it’s going to take over the world. Sam Altman says so.
>
> [image: 4992.jpg]
>
> Sam Altman’s make-or-break year: can the OpenAI CEO cash in his bet on the
> future?
> <https://www.theguardian.com/technology/ng-interactive/2026/jan/25/sam-altman-openai>
> theguardian.com
> <https://www.theguardian.com/technology/ng-interactive/2026/jan/25/sam-altman-openai>
>
> <https://www.theguardian.com/technology/ng-interactive/2026/jan/25/sam-altman-openai>
>
> You know the real problem with the enshitification of the internet is
> legal liability. If meta, OpenAI, Google, TikTok, x etc had the same legal
> liabilities as publishers, they would stop producing and distributing crap
> because otherwise they would be sued out of existence.
>
> Adrian
> Sent from my iPhone
>
> On 25 Jan 2026, at 16:24, Harry Powell <
> [email protected]> wrote:
>
>  I read this yesterday -
>
>
> https://www.theguardian.com/technology/2026/jan/24/latest-chatgpt-model-uses-elon-musks-grokipedia-as-source-tests-reveal
>
> And this (and similar articles) a while back -
>
> https://grokipediawiki.com/analysis/plagiarism-scandal-investigation/
>
> These don’t inspire me to use either.
>
> Harry
>
> On 25 Jan 2026, at 13:17, Hughes, Jon <
> [email protected]> wrote:
>
> hi,
> there has been much talk of using AI to write code for us and of it making
> the world better. people in this group have their own opinions regarding
> alphafold, for example, but at a much simpler level, i just asked chatGTP
> something about electrical power generation: his/her/their answer finally
> included, ""Interpretation per joule: 4–12 €cents per kWh equals 4–12 ×
> 10⁻⁶ € per joule, since 1 kWh = 3.6 MJ". well, we all make mistakes, right?!
> cheers,
> jon
>
>
> ########################################################################
>
> To unsubscribe from the CCP4BB list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
> This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a
> mailing list hosted by www.jiscmail.ac.uk, terms & conditions are
> available at https://www.jiscmail.ac.uk/policyandsecurity/
>
>
>
> ------------------------------
>
> To unsubscribe from the CCP4BB list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
>
> ------------------------------
>
> To unsubscribe from the CCP4BB list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>

########################################################################

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1

This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/

Reply via email to