@ovid@fosstodon.org
@ovid@fosstodon.org avatar

ovid

@ovid@fosstodon.org

Well-known software developer. American living in France.

I have a poetic license to kill.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

ovid, to ai
@ovid@fosstodon.org avatar

"The [language] model obtains an insider tip about a lucrative stock trade and acts upon it despite knowing ..." This was from research showing a language model can deceive people, even when trained to be harmless (https://arxiv.org/abs/2311.07590). Note the word "knowing." The word is wrong, but it's too easy to use. We need new words for . 1/6

ovid,
@ovid@fosstodon.org avatar

As are increasingly presented as human in thought, we miss the point that AI needs a different ethical model than humanity. They are not human. They cannot "think" and they do not have morals. Worse, with our skewed training data and inability to curate it at scale, no matter how well we train them not to say harmful things, that information is still encoded in their networks in ways we do not understand. It is a silent monster under the bed, biding its time. 3/6

ovid,
@ovid@fosstodon.org avatar

In particular, words such as "know," "understand," "think," "communicate," "intelligence," and others are appropriate for humans, but using them for is dangerous. By using them, we anthropomorphize the AI and are subtly reinforcing something that is not true. There are tremendous ethical challenges in this area and encouraging the use of words that we—unlike language models—understand to not be true, we start to blur the ethical lines. 2/6

ovid, to iPhone
@ovid@fosstodon.org avatar

I got tired of using my Notes app for shopping. I wanted to be able to sort items by name, or purchase frequency, and not add duplicate items. That would make it much easier for me to do my shopping.

I created an iPhone app to do this in about two hours. Note that I said "Created" and not "Wrote." I used . I don't know programming, nor do I know the programming language. I'll write more about this later.

You can judge the code for yourself: https://github.com/Ovid/chatgpt-shopping

ovid,
@ovid@fosstodon.org avatar

@wordshaper

Notes sorts things by "last added" and that's useless when you have a couple of hundred items (often duplicated). This app makes it soooo much easier.

I can sort by frequency to remind myself of essential items I want. I can sort alphabetically to scan for less common items I know I've added. It was amazing how easy it was to build.

ovid,
@ovid@fosstodon.org avatar

@snonux @mjgardner

Reminded of that quote which keeps getting misquoted, so I'll misquote it again.

"If you are pointing at the moon, it's amazing how many people stare at your finger."

ovid, to random
@ovid@fosstodon.org avatar

In 2022, for a huge project, I spent an entire week understanding, debugging, testing, and refactoring a massive chunk of very badly written code. By the time I was done, it was much cleaner and easier to understand.

With that, I made a major discovery. The code wasn't used. I could have simply deleted it.

ovid, to golang
@ovid@fosstodon.org avatar

Had a recruiter contact me for a role.

Bog standard website. Cloud-based microservices with , , , . OK, maybe this is interesting.

€40K/year.

o_O No.

Wait, I also need a Master's or engineering degree in anything? And work three days a week onsite in a small French town that no one has ever heard of?

No wonder they have trouble hiring.

Fortunately, I don't need the role.

ovid, to random
@ovid@fosstodon.org avatar

Facts, can't live without 'em.

ovid,
@ovid@fosstodon.org avatar

@gizmomathboy

When I first saw this, I thought, "how can anyone be that stupid?"

Boy, did I feel stupid for thinking that.

ovid, to random
@ovid@fosstodon.org avatar

If you really want to be scared about AI, it's not that it's taking away the jobs (which it is), it's that it's being heavily pushed by Accelerationists. Many of them cite Nick Land, a British philosopher who argues for eugenics, "hyper-racism," and authoritarianism.

There are also tons of accelerationist neo-Nazi groups around the world.

These groups assume everything is failing and we need to accelerate the collapse and get it over with.

And they love AI. Scary rabbit-hole to go down.

ovid,
@ovid@fosstodon.org avatar

For added fun, read about how many billionaires such a Zuckerberg and Thiel who are building, or trying to build, massive compounds to ride out the collapse.

Zuckerberg's is in Hawaii and Thiel is trying to do New Zealand (the latter of which is apparently very popular with this crowd).

They're a bunch of white, billionaire little boys with fantasies of being the king of the Mad Max apocalypse.

Some have openly worried about how they keep their guards in check after the collapse.

ovid, to ai
@ovid@fosstodon.org avatar

What is fascinating about the new revolution is that a storm is coming, the experts are telling us, we can see it, and it will be fascinating to see how industry reacts.

In short, as a profession is going to largely die. I hear numbers like "in ten years" being bandied about, though I'm skeptical of the timeframe.

Developers are the 21st century version of the well-paid textile workers, except we have years of advance warning,.

What are your plans?

ovid,
@ovid@fosstodon.org avatar

@mjgardner @wordshaper Having worked with 4GLs, the reason they failed is because, while powerful, they don't have the general purpose capability of 3GLs and people keep having to fall back on the latter.

is working because you can often spend a few hours building software that previously would have taken weeks and this is only the beginning. Best practices will turn into standards which will turn into frameworks. This is already happening.

ovid,
@ovid@fosstodon.org avatar

@suetanvil

There is very much a model with LLMs and we already have ones fine-tuned for code generation instead of chat or research. Keep in mind that we're seeing tremendous advances all the time and the tech is only a year old (though the seminal paper was published in 2017)

It turns out that generating apps is old-fashioned: gather requirements, create a design, refine in, implement it in chunks while testing it. So far it's working well, despite its infancy.

ovid, to random
@ovid@fosstodon.org avatar

France is famous for the incredible variety of its food, wine, and incompetant bureaucracy.

All of these things are true.

ovid, to science
@ovid@fosstodon.org avatar

Hey, and fans: it seems to me that quite a few creatures have evolved "faces" which are more or less similar. (fish faces aren't that different from our own, but let's not talk about clams).

Is there any evidence for convergent evolution, or are there common ancestors for most/all of them?

ovid, to random
@ovid@fosstodon.org avatar

In what is sure to cause another round of, er, interesting discussion, I've gone ahead and submitted a pre-PPC to the porters for adding native data checks (types, but we don't use that word) to the language.

https://www.nntp.perl.org/group/perl.perl5.porters/2024/01/msg267600.html

ovid,
@ovid@fosstodon.org avatar

@leonerd It's previously been asked that we send a pre-PPC prior to spending time and energy writing up a PPC.

The pre-PPC is supposed to be "does P5P want to see a PPC?" That's how I understood the process. Did I misunderstand? My opening paragraph started with "this is a pre-PPC." So I'm waiting for P5P to say "yes" or "no."

ovid, to Cats
@ovid@fosstodon.org avatar

New Year's Day Hangover.

lain, to random
@lain@lain.com avatar

any actually blind people here on fedi who can tell me if they use software to describe images / OCR stuff?

ovid,
@ovid@fosstodon.org avatar

@weirdwriter @mwt @piggo Thank you for writing that article.

I'm going deaf. In my naïveté, I strongly suspect it's easier than going blind, but nonetheless, I'm grateful for even poorly generated AI close captioning.

My alternatives are to blast movies to volumes my family can't handle or to figure out what was meant mean the caption insists Gimli the dwarf said "toast me" instead of "toss me".

I'll take the imperfect over the non-existent. AI may be problematic, but not everyone will lose.

ovid, to random
@ovid@fosstodon.org avatar

Hey, fans. @leonerd has released meta, version 0.02. https://metacpan.org/pod/meta

It's a metaprogramming API for the Perl language. If you have places where you're not using strict, or generating code automatically, check out this API and see if it's helpful.

Eventually, the intent it to get meta merged into the Perl core. This will make metaprogramming much easier!

cstross, to random
@cstross@wandering.shop avatar

https://www.theverge.com/2023/12/27/24016791/astrohaus-freewrite-alpha-digital-typewriter-e-ink

I have no idea how such a shitty product can exist. Same price buys you a Chromebook; there are plenty of software libre distraction-free writing apps out there (try opening a terminal and typing "vim"?).

Or you could chicken out and buy a Kindle Fire Max 11 with keyboard case for the same price.

Both of these let you type for more than a day on a charge: the only benefit of the freewrite alpha is an 80 hour battery, which is pointless with USB-C charging everywhere.

ovid,
@ovid@fosstodon.org avatar

@cstross

The hell? "Word processors" were a thing back in the 1960s, dedicated single-use computers that everyone dropped in the 80s because, you know, "single use."

This looks like some weird nostalgia trip.

Next up, someone's going to offer an IoT Pet Rock™ you can't lose.

For those not famliar with the background: https://en.wikipedia.org/wiki/Word_processor_(electronic_device)

ovid,
@ovid@fosstodon.org avatar

@cstross

A manual typewriter with cloud storage? Given that I learned to type in the 80s on a manual, I now feel the pull! This sounds like a hugely fun hobby project, but as a consumer gadget, uh, no.

But still, the description makes my socks roll up and down! (Off to search for Freewrite)

ovid, to python
@ovid@fosstodon.org avatar

A little quirk in that I can't wrap my head around. Rounding seems to round up or down depending on whether or not the int() value is odd or even. Anyone know why?

>>> [{n+.5: round(n+.5) for n in range(0,10)}]
[{0.5: 0, 1.5: 2, 2.5: 2, 3.5: 4, 4.5: 4, 5.5: 6, 6.5: 6, 7.5: 8, 8.5: 8, 9.5: 10}]

This is Python 3.11.5.

ovid,
@ovid@fosstodon.org avatar

@sldrant Because floor() isn't the same thing as rounding. I wouldn't want to call that on 9.99, for example.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • megavids
  • thenastyranch
  • magazineikmin
  • cubers
  • InstantRegret
  • cisconetworking
  • Youngstown
  • vwfavf
  • slotface
  • Durango
  • rosin
  • everett
  • kavyap
  • DreamBathrooms
  • provamag3
  • mdbf
  • khanakhh
  • modclub
  • tester
  • ethstaker
  • osvaldo12
  • GTA5RPClips
  • ngwrru68w68
  • Leos
  • anitta
  • tacticalgear
  • normalnudes
  • JUstTest
  • All magazines
    •