Note that 'human' = 'crowdworkers with minimal training', 'better' = 'better according to crowdworkers with minimal training'

Tbc I am *not* saying dehumanize annotators — platforms do plenty of that already. But we need to ask if this is an appropriate task for crowdsourcing

Scientific American Let's also be clear that
1. calculators are close to 100% reliable (unlike chatbots) and

2. we still teach kids arithmetic post-calculator; ergo we still need to teach kids writing post-bot. /7

Yes language and writing *are* cool and there *are* interesting things we can explore, but also we can (and should) do that without making up Weird Orientalist Nonsense™

In case you missed it, here's my article with @[email protected] on Mastodon introducing the TESCREAL bundle of ideologies, which has played--we argue--an integral role in shaping and driving the current race to build AGI.

firstmonday.org/ojs/index.php/…

Complaining about 'made up words' like i have news for you about all the words

find a way to hold your tongue until you stop renting

find a way to hold your tongue until you pay off your mortgage

find a way to hold your tongue until your kids are through college

find a way to hold your tongue until you retire

find a way to hold your tongue until you die

<3

Just today, I was asked how I felt watching people fall for AI hype. Exasperated. Exasperated is how I feel.

Can we just NOT?

Today is an excellent day to re-read this piece by Meredith Whittaker: logicmag.io/supa-dupa-skie…

Holy shit. Google fired 28 workers for protesting against its $1.2 billion contract with the IDF. Solidarity with these workers of conscience.

medium.com/@notechforapar…

We will not be following their lead.

My paper with @[email protected] on Mastodon on 'Envisioning Information Access Systems: What Makes for Good Tools and a Healthy Web?' is officially out in the ACM Transactions on the Web (TWeb) dl.acm.org/doi/10.1145/36…

'Like children clapping for Tinkerbell, we have to be optimistic because the grift dies if we’re not.'

For everyone who is using LLM driven chatbots as information access systems: this one is for you

Or, yes, better data matters. But we always have to remember that 'better' is being assessed via narrow benchmarks that DO NOT speak to efficacy or impact in the 'real world'

This is the correct response to 'if you oppose AI-powered X then you don't want people to have more access to X' or 'you clearly don't care about how overworked X providers currently are'

With IDF's 'Lavender' and its use in the violence in Gaza, the point is not the technology or its errors, but plausible deniability and ideologies of scale. New for the newsletter.

buttondown.email/maiht3k/archiv…

This is a historic tech action.

A long proud tradition. From Polaroid workers rejecting their work contributing to S Africa's passbooks, to the MIT research strike where scientists refused to work for warfare; to Geoff Hinton relocating to Canada to avoid building US weapons.