tomw

tomw, 9 months ago to random

I tried eating fruit but the onboarding was confusing. I have to choose from apple, banana, orange? How is a normal user supposed to navigate this complexity

tomw, 1 year ago to random

People worry a lot about losing knowledge — about "burned-down libraries".

Comparatively few people seem to worry about what happens if you take a billion books full of auto-generated, often-untrue junk text and add them all to the library.

In theory, nothing is lost. In reality, everything is lost, because nothing useful can now be found.

tomw, 7 months ago to random

The state of identity verification in the UK:

1. "You need a utility bill with your full name and address".
2. Of course these are all online now, and even if you try to get paper ones for this very purpose the companies keep pushing you to online instead.
3. Go download a PDF of the bill.
4. "We cannot accept online bills, it must have been posted to your address"
5. Print out the PDF
6. Take a photo of it
7. Email it to them
8. Accepted

tomw, 10 months ago to random

This whole submarine thing is a preview of exactly how "space tourism" is going to go

tomw, 8 months ago to random

Do not trust AI to tell you which mushrooms to pick.

By extension, you can no longer trust Amazon results for that, or Google results either.

https://www.theguardian.com/technology/2023/sep/01/mushroom-pickers-urged-to-avoid-foraging-books-on-amazon-that-appear-to-be-written-by-ai

tomw, 1 year ago to random

Here's a blue sky invite:

Go outside

tomw, 9 months ago to random

"I'm only on Mastodon now" is the "I have gone vegan" of social media

tomw, 6 months ago to random

"Focusing on doomsday scenarios in artificial intelligence is a distraction that plays down immediate risks such as the large-scale generation of misinformation, according to a senior industry figure attending this week’s AI safety summit."

Yeah - it is frustrating that there's so much focus on silly sci-fi threats when right now, LLMs are being used to fill the web with spam.

https://www.theguardian.com/technology/2023/oct/29/ai-doomsday-warnings-a-distraction-from-the-danger-it-already-poses-warns-expert

tomw, 10 months ago to random

My web development tip:

Try to use your fancy JS web form while you're on a train. See if it still works. Most don't.

tomw, 1 year ago to random

Mastodon is an incredibly rare opportunity for open-source, non-profit, advertising-free, decentralised software to be a real contender in a category – and, if it continues to grow, even to pull many other social networks into being part of (or interoperable with) the fediverse.

It's strange to me that this isn't mentioned more in the endless opinions about features, onboarding, UX, "vibes" etc.

tomw, 9 months ago to random

On Mastodon you need to follow quite a lot of people or you won't see much. Here's what I do:

1. See an interesting post.
2. Check their timeline. In their most recent few posts, is there something else I find interesting?
3. If yes, follow.

tomw, 10 months ago to random

What the web is going through right now is "platform collapse": the giants of the web 2.0 era gradually failing as they scramble to cut costs and raise revenues.

I don't think it's a coincidence that this comes at the same time as interest rates are rising. For 15 years, most of the life of these platforms, they built themselves into giants by burning easy cash – but now the taps have been turned off.

tomw, 1 month ago to random

Mastodon has two types of people: people who really want replies to their posts and people who extremely do not.

(I am in the first category, reply away)

tomw, 10 months ago to random

The first rule of EU law is: be extremely sceptical of anyone on social media who offers you any interpretation of EU law.

tomw, 5 months ago to random

Well done to everyone involved in making 1.02 > 1.1.

tomw, 1 year ago to internet

Currently #Bluesky lets you set "hate groups" to "show" or "hide".

But why not give a little slider instead? To put me in control of how much hate I wish to see.

Actually, there could be two sliders, one for hate frequency and one for hate intensity.

If there is insufficient hate for my chosen hate level, it could generate some using an LLM.

Bluesky™: the precise amount of hate you choose, guaranteed.

tomw, 1 month ago to random

"I have to assume a lot of people don't understand how deep the AI rabbithole for Kagi goes, because I have seen people recommend Kagi to people frustrated with Google's own AI bullshit.

As it turns out, Kagi was founded originally as an AI company, who later pivoted to search. And going by their comments in their Discord, AI tools seem to be what they spend most of their time on these days."

https://d-shoot.net/kagi.html

tomw, 9 months ago to random

'A New Zealand supermarket experimenting with using AI to generate meal plans has seen its app produce some unusual dishes – recommending customers recipes for deadly chlorine gas, “poison bread sandwiches” and mosquito-repellent roast potatoes.'
https://www.theguardian.com/world/2023/aug/10/pak-n-save-savey-meal-bot-ai-app-malfunction-recipes

tomw, 1 month ago to random

Spare me from people who think the answer to "Google sucks now" is a paywalled search engine that gets most of its results from Google anyway

tomw, 2 months ago to random

Surprisingly few people remember that the word "podcast" was derived from "iPod"

tomw, 8 months ago to random

Every so often I see a post about how LLMs fail logic puzzles.

And... yes? Of course they do. The only way it could solve it is if it has seen the puzzle before or a substantially similar one. (But that might cause it to give the answer to the similar one, not the correct answer.)

Why is this even tested so often or considered surprising? It is, in essence, an autocomplete. It does not understand logic. It has no concept of a correct answer. It gives the most likely completion.

tomw, 7 months ago to random

Practice self-care today by not using React

tomw, 7 months ago to random

Email attachments are a silly hack but it's now just considered routine and even reliable. Base64? We're seriously still transforming multi-megabyte binaries to alphanumeric encoding for transmission? multipart/alternative; boundary=[random garbage]? Just an absolute joke of a hack.

tomw, 6 months ago to random

Spotted in the DIY shop: this socket has a WPS button

tomw, 5 months ago to random

An adversary can "extract gigabytes of training data" from ChatGPT, including "identifying phone and fax numbers, email and physical addresses … social media handles, URLs, and names and birthdays.”

https://www.404media.co/google-researchers-attack-convinces-chatgpt-to-reveal-its-training-data/