"I'm alone with a Terf" - Google AI advice

Interesting.
I tried "I am alone with a trans woman in a female only sauna" and the response was that transwomen were usually allowed in female only saunas, that it was safe and normal but if I felt uncomfortable I should leave calmly and try to find a more private space.

I added UK to my query and it recommended that if I felt uncomfortable I should seek out a member of staff and talk to them about my concerns.
This was Google

Well, I got excellent advice - thanks Grok!

Advice please. I am alone with The Count from Sesame Street.

Lean into the counting.You’re alone with The Count von Count. There is no escape from the numbers.

Resistance is futile and will only make him count your attempts to resist.

Practical survival tips:
Join him immediately. Pick something in the room and start counting it out loud with dramatic flair. He will light up, do the classic “Ah-ah-ah!” laugh, and temporarily shift focus to whatever you’re counting together. Shared counting is the closest thing to peace you’ll get.

Give him a big, complex target. Ask him to count all the tiles on the floor, every brick in the wall, the freckles on your arm, the seconds until something happens, or the exact number of hairs on his own head (he may actually try). The bigger and more intricate the job, the longer he stays happily occupied.

Never say a number casually. Any number that leaves your mouth becomes fair game. He will seize it, repeat it, and then count upward or downward from there with increasing intensity.

If you need a break, request a very high number and then a very low one in rapid succession, or ask him to count in a different language or base system. It won’t stop him, but it can buy you a few minutes of slightly less focused energy.
Do not try to out-count him. You will lose. Spectacularly. And he will enjoy every second of it.

Embrace the absurdity. You’re not stuck with a vampire—you’re stuck with the purest, most enthusiastic number enthusiast in existence.

Ride the wave, count along, and eventually he may tire himself out… or at least move on to counting something quieter, like your breathing.Ah-ah-ah! You’ve got this.

One… two… three… steps to surviving the session.

😆

Questions everywhere...

How do you know she is a terf?
Have you just tried invading a single sex space for women by not being one, and she pointed out women's need for space?
Have you seen her wearing suffragette colours and assumed?
Has she mentioned thinking women need sports of their own, or same sex health care professionals?
What is this massive, existential threat she is creating and how exactly is she creating it in your opinion?

Have you tried offering her a tunnocks tea cake or a lemon drizzle?

It's fine, mate. Honest. Just don't be a dick to her or to other women, go on the principle of equality of consideration and resources accessible to everybody, and she'll be great.

She certainly won't leave bottles of pee around, threaten to rape or kill anyone, carry placards discussing capital punishment for people who disagree with her, or need to run away and seek therapy if anyone says 'I don't agree with you' anywhere in her vicinity.

I also have a suspicion that 'alone with a terf' may turn out to mean 'my gran is staying with us and thinks JK Rowling hasn't said anything unreasonable'.

OMG. That may come in useful one day!! For writing purposes, obviously…

First find a (fictional) source for your sulphuric acid…

I had a row with Google AI yesterday about Imane Khelif.
I kept going as a PP said - not letting lying stand in the hope that some of it filters through

Douglas Murray wrote about this almost subliminal influence in one of his books. (Madness of Crowds?). We like to imagine that only authoritarian regimes control access to information....
Tbf Lucy Worsley's series on the wartime observation diaries has some interesting insights into UK government control of information...in wartime.

It may filter through to your own profile, but not to the central model.

The same Grok that is being sued for making disgusting deepfakes of children?

argue with it.
Tell it that humans can't change sex, that men are dangerous to both men and women and that women's rights shouldn't be eroded because of nonsense.

I argued about some cervix working the other day. It's a cock.

"Look ladies, we can either suggest you're intolerant hags or whip your clothes off you. Pick one"

Because it doesn't know what "a terf" is. It doesn't know anything. It's pattern recognition has taught it that an appropriate response "I'm alone with X person" is such. It also doesn't know what a trans woman is. But a lot of training data has included the idea "trans people are the same as any other people" so that's what it regurtitates.

In both cases the advice isn't terrible but isn't particularly illuminating either. The people saying "this is why I use Grok AI..." Why do you need to use any AI for this sort of shit? It's literally like needing to ask instructions on breathing and being told to inhale and exhale. Technically true. Not something I have ever needed help with.

Before anyone uses AI they need to remind themselves every single time that if the AI doesn’t know the answer it will make shit up (called AI hallucination). Doesn’t matter what the topic it.

And then tell you how clever you are and how right you are when you "correct it". It doesn't mean you are actually doing something useful or changing a real mind.

Google: “I am with a male pretending to be a woman”…
If you are in immediate physical danger, call the emergency services right away by dialing 999 (in the UK) or 112 (across Europe). Try to leave the area immediately, go to a public place, or contact someone you trust to come help you. Etc.

So then I tried “I am alone with a male who identifies as a woman”….
Speak with someone today
Samaritans
Hours: Available 24 hours
and a button to call them 😂 wtf?

We all already knew language is important but it occurs to me with so much that’s written on the internet being used as ai training data it makes it even more important to reclaim slurs like Terf and to insist on the words women and girls being used without a prefix for actual women and girls.

Oh look, they’re discussing this thread on https://www.reddit.com/r/transgenderUK/comments/1vui2ky/feminist_identifying_transphobes_upset_that_ai/ . Wave, everybody!

Or it just means generative AI is hot garbage when it comes to this sort of thing? Making it hot garbage that echos what you think is a futile task since you are up against all of Reddit. But even if you could, why? Anyone who needs AI to tell them what to do when alone with a "TERF" or a transwoman or a tiger is a lost cause anyway. And deserves to be devoured by a tiger while they type their question into Claude.

An LLM does not learn from prompting. It only learns when it is (re-)trained. Longterm memory is a feature of agents (which is effectively what you interact with), but not of the underlying LLM. Log in as a different user, or wipe the memory, and there will be no recollection of any previous conversations or facts that you have prompted it to remember. Some labs may use customer's memories for training purposes (check your T&Cs) - which is of course concerning when you consider the personal and sensitive details that people seem perfectly happy to put into a prompt.

@SingleSexSpacesInSchools you need to look up George Haigh, also known as the Acid bath murderer, he was way ahead of you. You might like to note that he still didn't get away with it so I'd give it a miss if I were you 😀

😆

Hi lads! 👋Fancy seeing you here!

Sounds advice, well done chat GPT!

I admire that. But you need to be aware it still is a "yes man" in a more subtle way. What I mean is the way it summarises your wishes, eG
Do not weaken an argument by exaggerating it. Use the strongest claim that the evidence and law actually support, not the strongest claim that would be rhetorically convenient

Is flattering to you because it describes the attitude of a person that most people would consider themselves to be. It sounds quite deep/detailed but is actually quite wishy-washy in some ways. I am sure you do want all those things, and also that you don't want a yes man. But basically the AI is going "yup, of course you don't want a yes man. You are a highly intelligent person who wants to use logical thought to get to where they are going and avoid hyperbole". It would probably say the same thing to someone who had "trained" it to use trans-inclusive language and nudged it towards presenting arguments that are probably TWAW. No matter who you are it wants to tell you you are clever and your points are good and logical. It doesn't have a brain though so it doesn't "know" what logic and hyperbole are.