Dr H

Mapping climate policy: navigating context-specific language (Part I) — Climate Policy Radar In our latest blog, Anne Sietsma unpacks the mapping tool CPR developed to show how climate terminology differs across the globe.

We're opening up data on climate law and policy but context-dependent language makes it tricky. My colleague Anne Sietsma has an amazing blog post exploring this here www.climatepolicyradar.org/latest/mappi...

Graph showing inference costs of AI models over time, with the cost of GPT4 level intelligence falling 1000x in 1.5 years

Inference costs of GPT4 level intelligence have fallen THREE ORDERS OF MAGNITUDE in 18 months 😲😲

I've just switched over from CTO to Chief Architect at Climate Policy Radar! expect a bunch more posts on AI, knowledge graphs, esoterica, Satisfactory (I am crazy addicted.), and other things I now have a lot more time to get my head into. Looking forward to getting more tech in my life again

For everyone predicting 2025 as the year the ai bubble bursts, remember the market can remain irrational a lot longer than you can be solvent. And now we have artificial irrationality! There's still a huge amount of latent value to realise even if research progress stops dead.

Petition: An independent evaluation of the Cass review on child gender services We believe that trans healthcare should be based on unbiased research that is peer reviewed. We think that the Cass review's findings have led to restrictive practices that are being directly felt by ...

There's a parliamentary petition for an independent review of the Cass report into gender medicine for young people. It's a problematic document from by the previous government, shaped by TERFs.

If you're a Brit, you can sign it, whether in the UK or not.

petition.parliament.uk/petitions/70...

"while this outcome is not in line with our original research design, we do not feel it significantly impacts the central focus of this research" yea funny that wonder what incentives drove that conclusion

Sometimes academic sentences tell such a lovely story "Twelve students volunteered to participate in this study. Unfortunately, only one student completed all tasks" the perils of using students for research 😂😂

Anthropic derives 85% of its revenues from its API products vs OpenAI which is currently generating over 70% of its revenues from its chatbot premium subscriptions and only 27% from its APIs.

😲Aws hacking attempts up from 110m / day to 750m / day in 6months. Million!! The level playing field for skills genai gives people includes previously high friction crimes. How do you even begin to deal with that? (I feel like the answer is AI arms race...)

SG Report on the UN Disability Inclusion Strategy | United Nations The SG's report presents the system-wide efforts and collective commitments towards achieving transformative and lasting change for persons with disabilities within the Organization.

On International Day of #People with #Disabilities,
remember that some disabilities are #invisible
#Dyslexia
#Autism
#ADHD
#Dysgraphia
#Dyspraxia
#Dyscalculia
#disabled
#DisabilityRights
#DisabilityRightsAreHumanRights
#IDPwD
#IDPwD24
www.un.org/en/disabilit...

Combined with what chatgpt gives you when you ask "given what you know about me how could you manipulate me" I'm just waiting for the dystopian Cambridge analytica 2.0, hyper targeted manipulation feels scarily feasible

Trying to get perplexity to help debug nocobase, and it getting confused with nocodb sources in the search results, makes me think that coming up with good unique product names is even more important in the AI-search post-SEO era. Or risk users being given utterly perplexing (lol) info

With the members of his own family he kept up a regular correspondence, but with no one else. It was one of Chopin's peculiarities that he would write letters to no one else, and it might have been supposed that he had vowed he would never write to strangers. It was very curious to see him adopt all sorts of expedients to avoid writing only a short note. Many and many a time has he walked from one end of Paris to the other to decline an invitation to dinner or to afford some trifling information rather than write a line or two which would have saved all this trouble and waste of time.

Chopin would have had the *worst* possible phone and email anxiety. Relatable king

I would put this even more strongly: open source AI is probably our only realistic chance to avoid a terrifying increase in concentration of power. I do not want to live in a world where the people with all the money also have all the intellectual power.

Can I say something without people getting mad at me?

I don’t understand why one gets upset when people see the public posts on a public website, posted for people to see

I was supposed to go to a tech leaders dinner last night, but couldn't due to sensory overload. I <3 a lot of AuDHD, but this really made me feel disabled. like I can't access parts of the leadership world as sometimes my brain treats input as pain. How do we marry leadership and ND needs? #audhd

How We Make Sense of Time Long, short, forward and back: Our concepts of time—and how we process it in the brain—are based on our understanding of physical space, with some surprising cultural variations

www.scientificamerican.com/article/how-... also while finding this I realised that our brains intuitively use spacetime equivalence even though our physics for that only caught up with einstein

Anthropics product strategy is so on point rn I'm really loving it. Winning the hearts and minds of devs, but also QoL UX that openai's been neglecting. Proposing open protocols for ai/tool integration to undercut openai's by-default Microsoft customer base and level the playing field with llama. 🤌🤌

Dispelling myths of neurodiversity in the workplace - HR Magazine The neurodiversity awareness movement has been in full swing for a few years now, and has revolutionised our understanding of individual differences at work.

Dispelling myths of neurodiversity in the workplace:
www.hrmagazine.co.uk/content/comm...

#Autism
#Dyslexia
#ADHD
#Dyscalculia
#Dysgraphia
#Dyspraxia
📚
#Skybrarian

I think I was part of that. I was 6 at the time though so I don't feel that guilty

We built ours mostly manually using prefect to orchestrate, synthetic queries across various classes, and g-eval llm-as-judge eval approaches. Works great and turned up surprising insights. The real work was correlating evals against human judgements, but once comfy there very fun to scale!

It's getting cold in the house, time to open some Electron apps

The Onion should buy Elsevier next

my desk showing a computer on its side with a ryzen CPU tray and a mouse sitting on top as well as a monitor showing a bios update screen

everybody atheist till they gotta update a bios

This paper on legal personhood for AI www.yalelawjournal.org/forum/the-et... is one of the best reads this year imo. Wonderfully argued and resonating argument for me. + Dobbs leading to AI being afforded human-like rights would be dystopian irony to the max

Little device AI will be everywhere soon. Not super smart alone but weird together:
accidental car war for spots
Street cleaner convinces traffic lights to stay red so it meets goals
Customer bot measures better feedback when it dances
What happens when everything is a bit smart but also a bit dumb?

I recently learned that which direction you write in in your language is linked to how you see time and as a time blind ADHDer now I know why my writing is illegible

Sitting through streaming service adverts because you had to fast forward back to where the app crashed is basically charging us for their incompetence, it's so infuriating

Sometimes I look at a calendar and can't shake the feeling that time is just a neurotypical conspiracy designed to fuck with ADHDers