Olives @olives@qoto.org

Some articles are better, but there are enough of them which are just bad.

New challenge for The Guardian. Don't inadvertently apologize for fascists.

I'd be wary of ascribing fault to individuals as someone should be able to come clean about their mistakes without facing possible sanctions.

There are companies which you wouldn't expect which have made fairly bad mistakes.

While there are a few people blaming the company, one company having issues (CrowdStrike) shouldn't lead to so many having problems.

The problem is with so many companies (and others) relying on one particular service.

https://you.38degrees.org.uk/petitions/stop-the-met-police-using-facial-recognition-surveillance
"The Metropolitan Police has started using live facial recognition cameras across London. This surveillance technology turns us into walking ID cards and treats members of the public like suspects in a high-tech police line up. These cameras are a threat to our privacy and freedoms – and they need to be stopped.

The Met’s own stats show live facial recognition to be over 80% inaccurate and more likely to misidentify innocent people of colour as potential suspects. A court even found that police had been using this technology unlawfully."
#privacy #ukpol #FaceRecognition

There are upsides to this particular government, of course, but to put their worst decisions on a pedestal is really to betray one's own ignorance.

https://www.theguardian.com/commentisfree/article/2024/jul/22/taiwan-bucked-global-trend-trust-politics-hired-protesters From what I've heard, the "crackdown" on "deepfakes" has been an absolute disaster, with over-broad language posing a serious threat to freedom of expression, for instance, in "saving" imaginary people (#FreeSpeech). And this makes sense, as an average joe on the street is probably not that versed in policy.

It is so, so frustrating when someone from the U.K. voices stupid, stupid takes like this in complete ignorance of what is actually going on in the other side of world (#ukpol).

"When she left government, the administration’s approval ratings had reached highs of more than 70%."
Singapore has a ruling party with something like 90% of seats, and they execute people for drug trafficking. They are also against homosexuality.

If Facebook and Google are coming up with smart glasses, do you think the "glasshole" insult will return?

https://reason.com/2024/07/22/project-rescue-children-exposed-as-latest-in-series-of-human-trafficking-hoaxes/ British / Australian scammer grifting off fake child rescues. #ukpol #auspol

While I could see the other party also having this problem, one of the problems with Kamala is her historic stance against prostitution.

Even with something like privacy in interacting with the machine (or better yet, running it completely locally), that would be something to cover in a direct manner, not a doomsday manner.

If someone had written a concern that an algorithm might be used for disinformation, then well, that might be relevant (although, it is hard to say what can really be done about that). When someone talks like this, using this grandiose and dramatic language, and supposed future threats, it enters the realm of sensationalism.

https://www.theguardian.com/commentisfree/article/2024/jul/22/artificial-intelligence-panic-time-change
Notice the elements here. A sensationalistic headline. You should be afraid. Also, leaning heavily on tick tock tick tock, time is running out!, even though there is no sign as to why this week should be any different from any other week. People have made such claims over the past seven years (or even the past couple).

For good measure, they toss in a wad of *things you cannot possibly comprehend*, as if we are dealing with some kind of Lovecraftian horror veering over the horizon. This is a good example of "AI" sensationalism.

"If development continues at this speed"
This is not how "development" works. "development" is not made up of XP bars which you continuously fill up to unlock the "next level" of development.

https://reason.com/2024/07/19/the-doobie-act-would-limit-government-discrimination-against-people-who-have-smoked-weed/
"Sen. Gary Peters (D–Mich.) last week introduced a bill to prohibit the federal government from deeming an individual ineligible for employment or for a security clearance based solely on past marijuana use. The legislation is named the Dismantling Outdated Obstacles and Barriers to Individual Employment Act of 2024, otherwise known as the DOOBIE Act. Heh."

LLMs probably function better as a roleplay machine than as an assistant which is expected to be accurate.

Parts of the Australian Government questioning whether the PM's idea to ban minors from social media is a good idea.

It's interesting to see a take which says only 25 - 40% of data sources which could be tapped remain, and talking as if it is a disaster for "AI"*. I wonder if sucking up so much data was ever really sustainable though.

* I call it "AI" but I really mean artificial "intelligence". Writing the whole phrase out every time is tiresome, so I simplify it with a couple of quotes.

Ravi's takes on social media are very stupid and thoroughly out of touch with reality (unsurprising as he essentially parrots out of touch takes).

He suggests strong rate limits for fairly mundane things based on how a "typical user might use a site" on the basis that a *bad actor* might abuse a bit of functionality. This ignores though that there are *power users* who might make use of site functionality far more and there are *casuals*. It can even curtail the capacity of someone to block someone. This is how it pans out in practice. In reality, platforms have never provided unrestricted endpoints.

Ravi inadvertently suggests prohibiting fake porn and only allowing porn of real people to be posted. Clearly failing Welcome to the Internet 101. Curiously, he also presumes that content is always shared in a *broadcast* fashion, rather than casually with some friends (which is fairly common). He also demands a "fully moderated" social network where every post is carefully scrutinized by a company prior to being posted (impossible and harmful).

I wanted to take the time to cover this guy, because there is a small subset of out of touch people who like to "brainstorm" some really stupid ideas (completely out of touch with the real world), then talk as if they are "practical" things to do. People like this waste mental bandwidth from more serious discourse.