Will Beason

Reads: Most importantly, there is no AI without massive financial and ideological backing. It is therefore pointless to discuss its techniques or capabilities without asking who controls it, who benefits from it, who builds and deploys it, and what it is doing in the world. As Stafford Beer (2002) argued, the purpose of a system is what it does.

Reads: Though less explicit than Thiel’s call to replace politics with technology, major tech firms have effectively privatised core digital public goods. Platforms like Facebook, Google Search, and OpenAI’s ChatGPT operate at infrastructural scale in Ireland, shaping information, communication, and access to knowledge. Yet their algorithms remain opaque, their governance remains private, with minimal democratic accountability to the public who depend on them; effectively ceding aspects of democratic process to commercial interests. The monopolization of digital spaces has turned democracy into something the highest bidder can buy and is degrading the digital public goods themselves. As the AI industry, social media and search platforms grow more extractive and less trustworthy, they erode the foundations of democratic life: trust, dialogue, and accountability, blurring the line between truth and falsehood. An example is the deepfake video falsely showing President Catherine Connolly withdrawing from the presidential race last October, which amassed over 160,0001 Facebook views before being removed. GenAI’s non-deterministic, stochastic architecture produces plausible output without regard for accuracy or truth. This makes generative AI a societal disaster and a major threat to truth, democratic processes, information ecosystems, knowledge production, and the social fabric

Reads: For truth, democracy, and the rule of law to endure in the AI era, we need to cultivate an ecosystem of transparency and accountability. Yet governance by algorithms inherently places our digital public squares and democratic processes in the hands of those building these systems in line with their political and profit-seeking agendas. Without real mechanisms in place, talk of transparency and accountability are empty gestures. An internal Meta memo outlining plans to launch facial recognition in smart glasses “during a dynamic political environment where many civil society groups that we would expect to attack us would have their resources focused on other concerns”5 illustrates how those advocating for accountability are under-resourced, retaliated against, and targeted. Large tech and AI companies, despite selling promises of innovation and societal benefit, monetize and undermine the very society they claim to serve. What is needed is not just regulation, but active enforcement. Given the track record of tech giants, stricter regulation and enforcement is not “anti–freedom of speech” or anti-competitiveness. It is one of the clearest ways governments can show they serve the public interest. After all, innovation that disregards truth and democratic processes risks undermining democracy itself.

I appeared as an expert witness before the Joint Committee on AI at the Houses of Oireachtas (parliament of Ireland) to discuss "AI: truth and democracy" this morning. You can read my opening statement here: www.oireachtas.ie/en/publicati...

The problem with “AI will definitely take over programming” is that the hard part is not writing code. Fresh university students can usually accomplish that within a few weeks.

It’s writing code that does and keeps doing what it needs to do. And figuring out what that means. Over and over again. 🧵

Reminder that this tool was not for brain surgery but for sinus surgery, which is normally a relatively minor and safe procedure. I think some people in the QTs and replies are mistakenly thinking it was brain surgery due to the severity of the errors/outcomes, but it was just sinus surgery!

I didn't have TikTok on my phone when it was Chinese spyware, I'm not going to have it on my phone when it is US spyware

Also got a new 5k PR of 24:11!

Garmin view of 4.70 mile VO2 Max run. There is a personal record notification.

Distribution of heart rates in various zones. Most of the time was spent in Zones 4 and 5, and some in 3.

Garmin view of 1 hour strength workout and VO2 max run summary.

311 days to marathon. Difficult VO2 Max run followed by an hour of weight training. Going to feel it tomorrow.

Just worked out for two hours. Going to go eat an entire pizza.

Ever since the Stochastic Parrots paper was published, I've been fielding the question "How do I know that you're not (just) a stochastic parrot?" ... one that I find inherently dehumanizing, as I write about here:

journals-sagepub-com.offcampus.lib.washington.edu/doi/10.1177/...

Short 🧵>>

Garmin view of three walks, totaling about four miles.

Elevation chart showing nearly 400 foot gain.

I wasn't able to train yesterday, but had a nice walk up and down Queen Anne. Looking forward to incorporating that into my running routes.

Garmin view of Long Run of 7.43 miles at 9:26 pace around Lake Union.

316 days to marathon. Nice long run around Lake Union. The goal here was heart rate, which I am really liking.

Done with my first month of marathon training!

Garmin view of base run of 4.87 miles at 11:40 pace

Altitude graph of run by time, showing 400 foot climb up and down a hill.

318 days to marathon. Base run with a heart rate goal, so I took the opportunity to run up Queen Anne. It's neat how a heart rate goal keeps me at a steady workload even with hilly terrain. Also how fast my heart rate drops during inactivity - by 50 after a minute waiting for a crosswalk!

Garmin view of Anaerobic run. 5.44 miles at 9:13 average pace.

319 days to marathon. Difficult anaerobic intervals in the cold rain. I went to the Seattle Center since it's got nice, grippy asphalt for fast running even when wet. I'll probably go back for sprints - lots of straight open space where I can keep from running into people.

Emily Bender’s thread trashing The AI Doc is a must-read before you take other critics’ raves for this film at face value.

Me: *sick for a couple days; recovers*
Me: Okay, ready for training.
Garmin: I rescheduled your most difficult run from while you were sick.
Me: Uhhh, thanks. *does okay on run* At least tomorrow is easy.
<finally gets good sleep>
Garmin: You slept so well I made today's run EVEN HARDER!
Me: yayyy

Looking forward to the longer run tomorrow - it's my first with a heart rate target instead of a pace target.

View of two Garmin workouts, a threshold run of 37:08 averaging 8:23/mi and a bodyweight circuit workout.

320 days to marathon. Was sick for a couple days and rested from training. Today felt a bit harder - threshold running - but that makes sense since I wasn't feeling 100%.

YouTube video by LegalEagle The Execution of Alex Pretti Broke Me

youtu.be/nDEAWxG7Bq8?...

This past week I ran 26.2 miles in total, likely the first time in my life. If I stick with my plan, in a few months that'll be ~40 miles per week, which seems like it's the entry-level threshold for being able to "really" train for a specific marathon time (or races in general).

I (now) know it's a running proverb, but it's neat that just running slower and being deliberate about pacing is making my training quality improve considerably. Only running 7-mile loops was nice, but progress was slow and I couldn't just do that every day.

Garmin workout of 4.45 mile run at 9:09/mile pace

Garmin strength workout Core Circuit 2

323 days to marathon. A much shorter run than a couple days ago. Still feeling the longer run a bit, but not as much as I'd have expected.

The Garmin Appalachian Trail badge for tracking 3,500 kilometers on foot.

My first month of Garmin ownership at the beginning of 2024. 66.6 miles total.

The first year of me tracking steps via Garmin, Feb 2024-Jan 2025. 664.9 miles total.

The last 12 months of my activity on Garmin, Feb 2025-Jan 2026. 1,457.1 miles

I got my Garmin watch a little over two years ago to try and motivate myself to be more active. The first year I ran/walked over 600 miles. This second year I more than doubled it. Today I got the Appalachian Trail badge for tracking 3,500km (2,175 miles).

Longest I've run *continuously* though. The 15k I did was run/walk.

Garmin run view of 8.45 mile run at 9:08/mile pace.

325 days to marathon. Second longest run of my life so far - I assume that'll change quite quickly. It's nice that 9-ish minute miles feels like a pace I can sustain almost indefinitely.

Really liked There Is No Antimemetics Division. Finished in just a couple days; couldn't stop reading it.

I did 77 meters in 10 seconds during my best sprint!

Garmin workout view of sprint workout.

326 days to marathon. Hit a new max speed PR of 17.6mph while doing sprints!

I think I could still do better - need to find a real track, but the park in Cap Hill is pretty good.

I'd say "tamales" but it was actually just one big tamale.

Garmin view of 4.52 Threshold run and a 21 minute strength workout.

328 days to marathon. Difficult threshold run, but I pulled through and got a new 5k PR of 24:54! Followed it up with cross training. Now for a plateful of tamale!

i once read "the rights you think criminals should have are the rights you think everyone should have because the state can criminalize anyone" and it changed how i think about everything