bertil

20 ms·

211.
by bertil 3y ago
That’s presumably “Claude For (Google) Sheets” (I just spent more time than I care to admit trying to figure out why the Code for Sustainable Homes was relevant here.)
212.
by bertil 3y ago
If the motor is the expensive part, that makes sense. In a way, that’s still what some cheap crates of DIY equipment do: they have a drill, a sander, a vibrating polisher, etc. that are all extensions around a handle/motor core, and a
213.
by bertil 3y ago
I haven’t used their service, so I don’t know what you mean by “over-censor,” but there is a very large commercial opportunity for a model that can handle business tasks. Such models would benefit from being prudish and carefully asking que
214.
by bertil 3y ago
I’m curious if their lawyer has a defense in saying that they advised against it, but were told to try it anyway. I’m even more curious if the change of ToS alone could be grounds for a trial, even a class action—making the risk not even wo
215.
by bertil 3y ago
I have serious reservation about the system, but the key ones are: - It’s wrongly presented as a tool against fraud: it’s a tool to identify complex cases that are more likely to need human review. It is presented as unfair because files wi
216.
by bertil 3y ago
That reveals a major complication of the process that your partner was a part of: if you suspect content to be CSAM, you *legally must not* look at it, open it, etc. It’s a way to prevent bad people from abusing those positions, but at the
217.
by bertil 3y ago
> if Meta continues to roll out end-to-end encryption as planned, it would result in the loss of the vast majority of reports (92% from Facebook …) Wait: how does that prevent reports? The recipient presumably can see the message and sh
218.
by bertil 3y ago
There are different definitions of “end” in “end-to-end” * One being you communicate from your device to the servers securely; in that case, you can do keyword matching. If you try anything like that, a few engineers will insist Meta clarif
219.
by bertil 3y ago
It’s a summary of the public statement: > Today, Meta has chosen to rollout end-to-end-encryption on Facebook Messenger, which means they will no longer be able to keep children safe on their platform. > Today, our role in protecting
220.
by bertil 3y ago
Probably the most reasonable take, but it’s unclear how you would tell who is a child and who is not. I read your comment as a provocation to get the NCA to reveal that it’s not children they care about, which… I think some of them genuinel
221.
by bertil 3y ago
Klingon is definitely going to be in the top 50 languages covered…
222.
by bertil 3y ago
This is extremely common for every new technology: “upload,” “download,” “stream,” “google,” “FaceTime,” most code patterns, all the new ML apps, “venmo” or whatever the name of the app you use for payment, etc. all of those are taken as is
223.
by bertil 3y ago
Those two are based on previously known ambigraphs: * very closely https://www.pinterest.com/pin/giraffepenguin--13398215764267... * or directly inspired by, but the “young lady” prompt triggered the model to pick a dr
224.
by bertil 3y ago
Most of them are not “illusions” where you perceive two identical segments being different lengths because of tricks of human perception, they are ambigrams. They rely on humans’ ability to think of any three dots as two eyes and a mouth. T
225.
by bertil 3y ago
I’m curious if they could ask for permission from the original authors (who doesn’t love a fun puzzle?—and it’s not like the profit motive here is alarming): most licenses are default permission. You can always reach out and ask for a one-o
226.
by bertil 3y ago
I’d need to check, but if one set of “ear and hole” can be swapped with another set, both sets have to be identical in shape and color. But if they split and attach to other edges rather than swap, that creates further connection. If you th
227.
by bertil 3y ago
Not disagreeing with your statement (I think there’s a major misunderstanding around privacy and consent underlining all those arguments — and that is somewhat a personal failure) but: > Posters, signs and other such ads in the real worl
228.
by bertil 3y ago
I suspect it’s the lamppost effect: the data was here, so they used that rather than a more relevant observation that would have cost a lot more money. As it’s a health-related database, I suspect it’s a photo taken to control for acne, but
229.
by bertil 3y ago
I would look seriously into how that attractiveness was evaluated, and the measurement might be more accurate for men: 1. Criteria for men and women are very different: tall vs. thin is key. If the profiles list height but not BMI, that wou
230.
by bertil 3y ago
I'm a Monzo client and honestly, I'm surprised they even offer the option. They do everything on-line if they can: it's cheaper for them, and I find it more convenient, on top of all the reasons you listed. I suspect phone ca
231.
by bertil 3y ago
I don't think anyone assumes it's any different than what you describe: centralised, official server than let users authenticate. You might have wanted something else, but it's never been presented as a decentralised or open
232.
by bertil 3y ago
There are now non-bank alternatives with similar coverage. Freja is likely the most established provider.
233.
by bertil 3y ago
That's an argument to expand the system, no?
234.
by bertil 3y ago
Thank you so much! This is a fantastic resource that I somehow missed.
235.
by bertil 3y ago
It was proprietary when I made it, but no one seemed to care–there wasn't anything that wasn't painfully public in there. I'll probably edit and include it in a class I'm planning about teaching people about A/B tes
236.
by bertil 3y ago
Joke explainer: "Yi," "Er," and "San" are (the partial transcriptions in the Latin alphabet of the most common words for) "One", "Two," and "Three" in Chinese. I’m assuming that &#
237.
by bertil 3y ago
Are you saying that the Bellman equations already use the notation Q*, or are you saying that those equations (I’m not as familiar as I should be, sorry) are the obvious connection between the incoherent ramblings from Reuters? Because havi
238.
by bertil 3y ago
Giving AI space to represent and deliberate about the world separately (secretly, in the case of Diplomacy) is an obvious, but very productive step once “think step by step” has been established as a key improvement over LLM’s standard logo
239.
by bertil 3y ago
Or me when I’m playing Catharge at Diplomacy. I’m not good at this game, at all. To a point that scares my friends, actually.
240.
by bertil 3y ago
I don’t think that’s the case, but it would explain why the article is so bad. I genuinely have no idea what they are trying to do, but every detail is clearly wrong.

More ›