Kit Colliver's Avatar

Kit Colliver

@presentlyincorrect.bsky.social

I take my cats on walks, work on queering housing and social policy, take action where I can, and knit socks. ๐Ÿ“Edinburgh ๐Ÿข Research Associate, York Law School ๐Ÿณ๏ธโ€โšง๏ธ Organise with @rtie.org (@resistingtransphobiaedi on instagram) they/them

700 Followers  |  426 Following  |  193 Posts  |  Joined: 17.11.2024  |  2.5255

Latest posts by presentlyincorrect.bsky.social on Bluesky

Does body doubling help? Thereโ€™s a zoom writing group I use with morning and afternoon sessions www.szwg.co.uk

07.10.2025 12:37 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Why Language Models Hallucinate, by Kalai et al. 

Like students facing hard exam questions, large language models sometimes guess when
uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such
โ€œhallucinationsโ€ persist even in state-of-the-art systems and undermine trust. We argue that
language models hallucinate because the training and evaluation procedures reward guessing over
acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern
training pipeline. Hallucinations need not be mysteriousโ€”they originate simply as errors in binary
classification. If incorrect statements cannot be distinguished from facts, then hallucinations
in pretrained language models will arise through natural statistical pressures. We then argue
that hallucinations persist due to the way most evaluations are gradedโ€”language models are
optimized to be good test-takers, and guessing when uncertain improves test performance. This
โ€œepidemicโ€ of penalizing uncertain responses can only be addressed through a socio-technical
mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate
leaderboards, rather than introducing additional hallucination evaluations. This change may
steer the field toward more trustworthy AI systems.

Why Language Models Hallucinate, by Kalai et al. Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such โ€œhallucinationsโ€ persist even in state-of-the-art systems and undermine trust. We argue that language models hallucinate because the training and evaluation procedures reward guessing over acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern training pipeline. Hallucinations need not be mysteriousโ€”they originate simply as errors in binary classification. If incorrect statements cannot be distinguished from facts, then hallucinations in pretrained language models will arise through natural statistical pressures. We then argue that hallucinations persist due to the way most evaluations are gradedโ€”language models are optimized to be good test-takers, and guessing when uncertain improves test performance. This โ€œepidemicโ€ of penalizing uncertain responses can only be addressed through a socio-technical mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate leaderboards, rather than introducing additional hallucination evaluations. This change may steer the field toward more trustworthy AI systems.

Ironically, it appears that AI chatbots hallucinate for the same reason that students feel compelled to use them:

They were socialized in a high-stakes testing culture that rewards guessing and maybe getting it right over admitting when there's something you just don't know.

08.09.2025 10:41 โ€” ๐Ÿ‘ 1460    ๐Ÿ” 414    ๐Ÿ’ฌ 43    ๐Ÿ“Œ 57

For my brotherโ€™s kids, I am Pibl Kit (their parentโ€™s sibling) and they are my niblings.

28.08.2025 13:51 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

How much did this failed experiment in social cleansing cost, we wonder? God forbid they engage in meaningful consultation before titting about like this.

28.08.2025 08:18 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Itโ€™s an important and exciting project and the contract duration and scope for independence are rarely seen in the world of fixed term research positions.

I donโ€™t know the PI personally but my partner speaks very highly of her.

26.08.2025 08:27 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

This looks rather wonderful!

26.08.2025 08:25 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Youโ€™re brilliant, Ruth. So well-deserved.

04.08.2025 17:15 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Reading this back and boy am I excited about Ruthโ€™s zines and sexual health.

10.07.2025 19:09 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Itโ€™s here!

ruthpearce.net/2025/05/26/p...

Also the folks at that drop-in are lovely. Get tested! In person! All the ones you actually need! No appointments!

10.07.2025 19:08 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Lolilol Iโ€™ll omilil milk lmikmll Iโ€™ll il Iโ€™ll lm lam zzi

10.07.2025 15:28 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

๐Ÿ“Œ thank you

03.07.2025 19:42 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Gratifying to see a legal view line up with my own instinctive interpretation. @reactiveashley.bsky.social is arguing that the Supreme Court ruling is consistent with implementing a โ€˜bigots toiletโ€™ and retaining trans-inclusive single-sex facilities.

01.07.2025 07:38 โ€” ๐Ÿ‘ 7    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Lemkin Institute for Genocide Prevention 
Red Flag Alert on Anti-trans and intersex Rights in the UK
Released on 30th of June 2025

Lemkin Institute for Genocide Prevention Red Flag Alert on Anti-trans and intersex Rights in the UK Released on 30th of June 2025

All of the actions described above fit neatly into the 9th Pattern of Genocide: "Denial and/or Prevention of Identity." As we have repeatedly stated over the years, genocide does not only manifest in the killing of an entire group. In the case of trans and intersex people, genocide is often perpetrated by making it impossible for individuals to exist as their true selves. The erasure of a group from public life is a step towards an attempt to erase that group's existence, which is the very definition of genocide. The UKSC's decision puts trans and intersex people at peril by forcing them to either out themselves or break the law, both of which run the risk of violent backlash. To explicitly confirm this, gratitude was extended in the UKSC ruling Ben Cooper KC, acting on behalf of Sex Matters. The latter is an explicitly anti-trans lobbying group that has since written to the NHS claiming that transitioning "becomes a major, permanent social problem for those (transgender) young people" โ€” a claim that flies in the face of all serious research on transgender youth.

All of the actions described above fit neatly into the 9th Pattern of Genocide: "Denial and/or Prevention of Identity." As we have repeatedly stated over the years, genocide does not only manifest in the killing of an entire group. In the case of trans and intersex people, genocide is often perpetrated by making it impossible for individuals to exist as their true selves. The erasure of a group from public life is a step towards an attempt to erase that group's existence, which is the very definition of genocide. The UKSC's decision puts trans and intersex people at peril by forcing them to either out themselves or break the law, both of which run the risk of violent backlash. To explicitly confirm this, gratitude was extended in the UKSC ruling Ben Cooper KC, acting on behalf of Sex Matters. The latter is an explicitly anti-trans lobbying group that has since written to the NHS claiming that transitioning "becomes a major, permanent social problem for those (transgender) young people" โ€” a claim that flies in the face of all serious research on transgender youth.

In the time itโ€™s taken me to write up my research on the impact of voter ID on trans people, the UK has managed to slide from a position of impinging on trans citizenship rights to setting off alarm bells at anti-genocide organisations.

01.07.2025 06:20 โ€” ๐Ÿ‘ 11    ๐Ÿ” 5    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

My overall impression is that the EHRC guidance does not make it clear how to deliver trans inclusive services.

It is effectively advice on how organisations can exclude trans and gender diverse people from services, which is an unusual approach for a human rights organisation to take.

30.06.2025 21:48 โ€” ๐Ÿ‘ 47    ๐Ÿ” 10    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 1

Spent the entire evening writing up the EHRC response for @rtie.org

And thatโ€™s with membersโ€™ feedback already collated.

That guidance and the consultation are trash - lengthy, confusing, mired in bigotry - and I feel like my eyes are bleeding.

30.06.2025 21:34 โ€” ๐Ÿ‘ 2    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Graphic with a purple background. In a bright green text box at the top it says "The EHRC's Consultation on their draft Code of Practice closes today!". Underneath are three drawings of placards, they say "Trans people belong", "Inclusion not Exclusion" and "No segregation! Trans Liberation". The Scottish Trans and Equality Network logos are at the bottom.

Graphic with a purple background. In a bright green text box at the top it says "The EHRC's Consultation on their draft Code of Practice closes today!". Underneath are three drawings of placards, they say "Trans people belong", "Inclusion not Exclusion" and "No segregation! Trans Liberation". The Scottish Trans and Equality Network logos are at the bottom.

Its the final day to complete the EHRC's consultation on their draft code of practice. If you want to fill it in and share your voice in support of trans inclusion, todayโ€™s your last chance!

We've created a full guide w/ Mermaids & TransActual, on our website - www.scottishtrans.org/supreme-cour...

30.06.2025 08:02 โ€” ๐Ÿ‘ 86    ๐Ÿ” 65    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 5
Meme image of Bernie Sanders, with text that reads, I am once again updating my mega post about how awful the Cass Review is.

Meme image of Bernie Sanders, with text that reads, I am once again updating my mega post about how awful the Cass Review is.

I've added FOUR new peer-reviewed academic articles to my round-up post, "What's wrong with the Cass Review?" Together, these highlight serious ethical issues with the Cass Review's design, recommendations, and implementations. It's important to keep this story live! ruthpearce.net/2024/04/16/w...

28.06.2025 08:41 โ€” ๐Ÿ‘ 338    ๐Ÿ” 149    ๐Ÿ’ฌ 12    ๐Ÿ“Œ 3

I feel validated.

13.06.2025 11:06 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
From bathrooms to the ballot box: UK trends undermining trans* citizenship - Transforming Society Ash Stokoe and Kit Colliver argue that, amid intensifying anti-trans rhetoric and legal rollbacks, trans* people in the UK are increasingly excluded from public life, with voter ID laws and a recent S...

For Pride month, @ashstokoe.bsky.social and I have a blog in Transforming Society, 'From bathrooms to the ballot box: UK trends undermining trans* citizenship'.

We're looking at the Supreme Court's 'sex' ruling in light of our voter ID research. www.transformingsociety.co.uk/2025/06/11/f...

12.06.2025 08:52 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Some kind of crafting - I knit - and audiobooks! People suggested podcasts to me when I had the Big Concussion, but I didnโ€™t have enough exec function to make choices. Shout if youโ€™d like recommendations!

11.06.2025 07:41 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Porty Pride and Resisting Transphobia in Edinburgh logos above a beach background.
Text:
RTiE Big Trans Picnic
Join us on the beach for a relaxed gathering of trans people and their allies. We're here to find new friends, reconnect with familiar faces and build community together. 
All welcome.
Saturday 7th of June
13:00-15:00
Cake Stand on The Beach
49 Promenade, EH15 2BA

Porty Pride and Resisting Transphobia in Edinburgh logos above a beach background. Text: RTiE Big Trans Picnic Join us on the beach for a relaxed gathering of trans people and their allies. We're here to find new friends, reconnect with familiar faces and build community together. All welcome. Saturday 7th of June 13:00-15:00 Cake Stand on The Beach 49 Promenade, EH15 2BA

Big Trans Picnic @ Porty Pride 07 June 1-3pm!
Bring blankets, food, water and sunscreen to share or just for yourself, well behaved children and badly behaved pets. This is an outdoor event.
If the rain gets bad we will use our contingency venue @ Porty Town Hall - look out for our stories!

06.06.2025 18:03 โ€” ๐Ÿ‘ 8    ๐Ÿ” 5    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
A tall white woman in a floral short playing saxophone in front of a PowerPoint presentation. Black text on a white background reads: The data was hard to gather, but it tells a story about whโ€ฆ itโ€™sโ€ฆ trans people whose gender is recategorisedโ€ฆ thereโ€ฆ a way to tell those stories.

A tall white woman in a floral short playing saxophone in front of a PowerPoint presentation. Black text on a white background reads: The data was hard to gather, but it tells a story about whโ€ฆ itโ€™sโ€ฆ trans people whose gender is recategorisedโ€ฆ thereโ€ฆ a way to tell those stories.

The Queer Data showcase is fantastic. Lots of excellent funny, thought-provoking presentations and my highlight so far has been
@katherinewyers.bsky.social with improv sax accompanying her story of the emotional toll of queer fieldwork.

05.06.2025 17:27 โ€” ๐Ÿ‘ 11    ๐Ÿ” 4    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 2

A single sociologist. Not a medic. Not even an expert in gender and sexuality. A guy whose academic contribution and teaching is on social movements and protest actions, gone entirely off-piste for the sake of bigotry.

02.06.2025 18:13 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Thanks for all the work on it. Unfortunately the link youโ€™ve posted isnโ€™t working?

02.06.2025 18:07 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

๐Ÿ“Œ

02.06.2025 15:59 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
A map of the UK coloured according to where Iโ€™ve lived (5 points), stayed (4), visited (3), stopped (2) or passed through (1). A lot of it is coloured orange, meaning Iโ€™ve stayed there. The total score is 245.

A map of the UK coloured according to where Iโ€™ve lived (5 points), stayed (4), visited (3), stopped (2) or passed through (1). A lot of it is coloured orange, meaning Iโ€™ve stayed there. The total score is 245.

Data collection has sure gotten me about a bit. smstone0.github.io#/uk-map

31.05.2025 15:04 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 1
A hand-knitted sock in grey, black and brown against a coral coloured wall.

A hand-knitted sock in grey, black and brown against a coral coloured wall.

Light grey tracksuits tucked into two hand-knitted socks in grey, black and brown, photographed against a coral coloured wall.

Light grey tracksuits tucked into two hand-knitted socks in grey, black and brown, photographed against a coral coloured wall.

One sock. Now two socks.

28.05.2025 19:23 โ€” ๐Ÿ‘ 5    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Thereโ€™s been murmurs of setting up some online silent writing sessions through the trans scholars network. Iโ€™d love to see that - and any other plans for writing with company - come to fruition.

27.05.2025 17:13 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

I hear you!

25.05.2025 09:25 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Fab, thanks Chris! Wonder if thereโ€™s merit in an emergency amendment on that front - but I donโ€™t know enough about congress timeframes!

25.05.2025 09:22 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

@presentlyincorrect is following 19 prominent accounts