I know there are differences between male & female anatomy, but…


This is ridiculous.

Backwards, grossly mislabeled, multiplication of organs — two vaginas in unexpected places? — one imaginary organ (dorsme?), all professionally illustrated. Don’t show this to any young boys, they will be even worse in their future interactions with women.

Comments

  1. Silentbob says

    Okay, so we’ve figured out dames have three rectums, um recta, and two anuses, one of which is on the sole of the right foot – but still nobody can find that bloody clitoris! Keep searching lads.
    X-D

  2. Silentbob says

    (Waiting for Morales to turn up and tell us AI is brilliant actually were just too inferior to get it X-D )

  3. lotharloo says

    @Silentbob:

    If you discount things such as AI solving millennium math problems, yeah, AI is absolutely useless.

  4. M'thew says

    Why would AI bother to learn about human anatomy if it’s going to exterminate us in the not-too-distant future?
    I wonder if it will develop a sense of humour and build Daleks to EXTERMINAAAAAAATE us all.

  5. says

    @4 lotharloo said: ‘things such as AI solving millennium math problems’
    However, it is not that simple:
    https://tech.yahoo.com/ai/articles/openai-says-solved-maths-deepest-021600105.html
    OpenAI Says It Solved Math’s Deepest Problem But Mathematicians Say AI …
    5 days ago OpenAI says its AI solved a $1M math problem, but two mathematicians say it used their unpublished work to get the solution.
    and
    https://inshorts.com/en/news/mathematicians-skeptical-of-openai-s-navier-stokes-proof-claim-1788961305632
    Mathematicians skeptical of OpenAI’s Navier-Stokes proof claim
    5 days ago OpenAI claims its AI has generated a proof for the NavierStokes equations, one of the Millennium Prize Problems. Mathematicians remain skeptical because the document has not been made public. Experts say the claim requires rigorous independent scrutiny to determine whether the proof is

    SO: is quite possible that a.i. is not a genius, it’s just a thief with great derivative computation capability.

  6. lotharloo says

    @shermanj:

    Probably they did not. But it doesn’t matter because the said mathematician himself used AI. It also doesn’t matter because AI has already one-shot many major mathematical conjectures such as double cycle cover conjecture, Erdos unit distances problem, as well as making incredible progress on other problems. I think AI is already at a level above at least 50% of mathematicians/theoretical computer scientists, meaning, it can prove everything they are capable of proving.

  7. stuffin says

    grossly mislabeled is an understatement.

    Abdominal organs in the thoracic cavity, Two anuses, two appendixes and two vaginas. Well, the last one should improve my chances of finding that clitoris.

  8. Reginald Selkirk says

    @9

    Good point. AIs are also good at spreading rumors. They are a gold mine for propagandists.

  9. Gaz Smith says

    One of those great mysteries of life has finally been explained. The positions of the two vaginas explain why 50% of men like women with big bums and the other 50% have a foot fetish

  10. bcw bcw says

    It would be really hard for a human to make this, since humans start with basic structural understanding and fill in the details, while AI understands nothing but has millions of pictures of local relationships to smear together. It’s kind of like doing a Mr Potatohead with pieces drawn without looking at them and fit into place.

  11. Ridana says

    So the hip bone is connected to the vagina bone! And from that illo, apparently vagina spurs are a possibility. Maybe Dumpster wasn’t faking that after all.

    Although not all are labeled, I think it’s safe to assume that women have four vaginas, one on each hip, and one on each heel.

  12. says

    To lotharloo and CompulsoryAccount7746, Sky Captain; To be clear, I don’t doubt that a.i. is appropriate for these computational tasks (and some heavy stat-analytical tasks). However, as the mathematicians point out, to help prevent wild, unsubstantiated, obfuscatory claims, a.i. must be transparent, ‘show its work’ and must be able to be fact checked by humans.

  13. CompulsoryAccount7746, Sky Captain says

    “¿Cuántos anos tienes?”
    Google and Bing Translate refuse to acknowledge that omitting the tilde makes it a different word with a different meaning, even in isolation. A classic grade school giggle ruined. Admittedly an uncommon plural.

    Wikipedia – Syllis ramosa

  14. says

    Just one more comment and then I must stop and not hog the arena.
    @24 CompulsoryAccount7746, Sky Captain wrote: “¿Cuántos anos tienes?” and referenced a worm with one head and numerous anuses.
    I reply: You just almost described the tRUMP administration. However, the head of that administration is also an anus!

  15. lotharloo says

    @CompulsoryAccount7746, Sky Captain:

    Sure but that’s not what the discussion is about. I replied to a comment that claimed “AI is useless”, and OP is basically a post in similar vein, i.e., stupid AI doesn’t even know women don’t have vaginas in their ankles.

    I have no doubt openai has burned millions of dollars trying to prove those which to be frank is peanuts if you measure the many thousands and probably tens of thousands of mathematicians who have also tried to solve the same problem.

    Other conjectures have been solved pretty easily. Proving the double cycle cover conjecture probably cost a few hundred dollars. The proof is 3 pages: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98d31/cdc_prompt.pdf
    And here’s the prompt: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98d31/cdc_prompt.pdf

  16. lotharloo says

    @23 shermanj:

    The proofs are followed up by lean proof verification. Basically, what lean does is to check that all the steps of the proof are sound and they follow from principles. Once the lean verification is passed, the only thing left for the humans to check is to verify that the statement of the theorem is correctly given in lean. So once you have the lean verification, there is really not much to do just to check the statement in lean.

  17. says

    This image illustrates one thing perfectly; whatever generated this image and labels lacks any understanding of what it produced.

    It’s just like it mashed together every anatomical image ever produced.

  18. CompulsoryAccount7746, Sky Captain says

    lotharloo @28: “not much to do just to check the statement in lean.”

    MIT – “Cryptography and Machine Learning” lecture notes (pdf)

    Lean verifies proof terms against theorem statements. It has no access to mathematical intent. The following failure modes all live entirely outside the kernel’s reach:
    * Wrong object. Did I define the right mathematical structure? A subtly incorrect definition may admit a valid proof of the stated theorem while being meaningless or misleading for the intended application.
    * Wrong statement. Did I state the theorem I intended? A statement can be formally provable yet subtly different from the desired result.
    * Forgotten hypothesis. Did I accidentally omit a side condition? The theorem may be true only in a special case, but the formal statement is more general than intended.
    * Totalized APIs hiding boundary cases. Lean’s standard library often totalizes partial functions—defining them to return a default value (typically 0) where they would otherwise be undefined. A theorem stated using a totalized API may appear to have no hypotheses while silently sweeping boundary cases under the rug.
    […]
    an LLM may not be explicitly trying to deceive, but will find workarounds to produce compiling Lean code even when those workarounds are unsound

    Reiterating this from the other thread.
    Leiden Declaration on Artificial Intelligence and Mathematics

    Current automated techniques can produce plausible but unreliable (or even incorrect) arguments which are difficult to distinguish from correct mathematical proofs. This applies not only to informal arguments, but also to formalizations, where the difficulty lies in the translation between computer-encoded and human presentations of concepts.

  19. John Morales says

    This image illustrates one thing perfectly; whatever generated this image and labels lacks any understanding of what it produced.

    Not whatever; whoever. LLMs don’t spontaneously create images without prompting.

    The grotesqueness is deliberate, and it’s been making the rounds.
    No credited author, but a reverse search shows it was created early this year, and quickly spread on social media (particularly facebook and instagram) and popularised by AI-hating groups (e.g. Facebook Cursed AI Group).
    Probably generated for engagement-farming sites that use AI graphics to churn out daily content and clickbait.

  20. John Morales says

    CA7746, you keep missing the point.

    Current automated techniques can produce plausible but unreliable (or even incorrect) arguments which are difficult to distinguish from correct mathematical proofs.

    Two points: ‘can’ ≠ ‘will’, and ‘difficult to distinguish’ means it is difficult for humans to keep up.

    (well, one more: humans can also produce plausible but unreliable (or even incorrect) arguments which are difficult to distinguish from correct mathematical proofs :)

    Shall we trade quotations?
    Here: https://arstechnica.com/ai/2026/06/openais-math-breakthrough-played-to-ais-strengths/

    In mid-May, OpenAI announced that an internal AI model had disproved the Erdős unit distance conjecture, a famous problem in discrete geometry that had stumped human mathematicians for the last 80 years.
     
    OpenAI gave several mathematicians early access to the result and published their reactions. Tim Gowers—who won the Fields Medal, the most prestigious prize in mathematics—wrote that “there is no doubt that the solution to the unit-distance problem is a milestone in AI mathematics.”

  21. francesconic says

    27 lotharloo

    No you made the comment about AI is useless and then replied to it. You created a straw man to argue against. No comment before yours said AI is useless.
    One could hypothesise that you were trying to derail the argument being made which is something like , much of current AI is not fit for purpose.

  22. John Morales says

    francesconic, you get you are conceding the point?

    much of current AI is not fit for purpose.

    Interesting phrasing.
    That is equivalent to saying some of current AI is fit for purpose.

    This may amuse you: https://www.abc.net.au/listen/programs/lawreport/the-rapid-rise-of-ai-use-in-employment-law-disputes/107079374 — article is 1 hour ago, about an academic who knows computing but not lawyering.

    A Sydney computing academic made headlines when he won a landmark case at the Fair Work Commission by relying entirely on artificial intelligence.

    The Law Report explores concerns over the rapid rise in the use of AI by self-represented litigants in employment law disputes.

  23. Silentbob says

    @ wsierichs
    Yes, you’re absolutely correct and one thing that’s strikes you when you go back and look at the original episodes is how groundbreaking they were in the context of their time.

    In the first season episode “Naked Time” the premise is the crew are exposed to a contamination that makes them uninhibited and hallucinating.
    Spock gets the infection and collapses in tears. He’s incapacitated. Eventually Kirk asks what’s wrong and Spock says, “You don’t understand Jim… on Vulcan it’s shameful to show emotion… I could never even tell my mother I loved her!” And collapses again into a blubbering mess.

    This relevation of Spock’s inner life reportedly immensely boosted the character’s fan mail, but my point is: This was not mindless entertainment. This was not Rocky Jones. This was tackling serious issues (in this case toxic masculinity) in a sci-fi format. So many first season episodes are like that. And this was 1966. X-D

    I never even learned the term “toxic masculinity” for about another 40 years, but Star Trek was on it back in the day.

  24. beholder says

    @31 John Morales

    Not whatever; whoever. LLMs don’t spontaneously create images without prompting.

    The grotesqueness is deliberate

    Indeed. It’s so lazy to post these images without also posting the prompt. Chances are this was the intended output — make it look like an anatomy diagram, wrong answers only, etc.

  25. StevoR says

    @35. Maybe the A was depicting Vulcan anatomy instead of human anatomy? Missing the pointy ears & green blood but then the top of the head appears cut off so maybe they’re there just not in this pic? Mind you, I don’t recall any of those, umm, peculiar anatomical features being mentioned regarding Vulcans in any of the shows and series.

  26. birgerjohansson says

    Beavis and Butt-head thought “bisexual” meant “having two schlongs”. This image shows what happens when you train AI on 90s television.

Leave a Reply