@bornach@masto.ai avatar

bornach

@bornach@masto.ai

I'm an ex-postdoc researcher who was bullied out of academia over a decade ago

I now pursue my interests in
#science #technology #education #art #mathematics

via online content creation that explores ideas in #computer #programming, science #communication, #visualization, #electronics #circuit design, #cardboard #crafts, kinetic sculpture, #synthesizer music, and machine learning

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

timnitGebru, to random
@timnitGebru@dair-community.social avatar

Via @tomstoneham

"All this seems to be based on an article of faith; namely, that all that is needed to create superintelligent machines is (a) infinitely more data and (b) infinitely more computing power. And the strange thing is that at the moment the world seems to be taking these fantasies at face value."

https://www.theguardian.com/commentisfree/2024/apr/13/from-boom-to-burst-the-ai-bubble-is-only-heading-in-one-direction"

bornach, (edited )
@bornach@masto.ai avatar

@PJ_Evans @timnitGebru @tomstoneham
Won't stop them from trying

They're gambling that the $100B "Stargate" AI compute resource will either give them AGI, or earn/win enough investment to build the $1T successor. Rinse and repeat.
https://www.forbes.com/sites/cindygordon/2024/03/31/microsoft-and-openai-partnering-on-stargate-a-100b-us-data-center/

To get enough data to feed the thing, they overhype AI to convince companies to layoff staff. Unemployed highly-skilled workers are forced into the gig economy doing data labelling tasks for AI companies. Rinse&repeat
https://youtu.be/vQChW_jgMMM?t=13m24s

bornach, to ai
@bornach@masto.ai avatar

[ColdFusion] on the rising trend of AI Washing -- overhyping AI and whether we are in an AI bubble.
https://youtu.be/vQChW_jgMMM

bornach, to ai
@bornach@masto.ai avatar

I am shocked, shocked to find out that AI is being overhyped by start-ups seeking investor funding and by influencers looking for clicks
https://youtu.be/tNmgmwEtoWE

futurebird, (edited ) to random
@futurebird@sauropods.win avatar

People trying to train AIs are now complaining that all of the AI data on the internet are making it hard for them to get quality training sets of natural language and images.

bitter snickering

bornach,
@bornach@masto.ai avatar

@futurebird @millerdl
So AI LLMs won't know of the Will Smith slap meme, that a British PM had the shelf life of a lettuce, that a senior trial lawyer in the UK is now called a King's Counsel, or that Twitter has become a letter of the alphabet where blue checkmarks are to be disparaged.

That might be useful for detecting AI bots.

bornach,
@bornach@masto.ai avatar

@cehteh @futurebird

It seems like started marking their own generated text a few months ago. One of the telltale red flags is curiously whether the article contains an out of context fun fact about Mount Everest being 46,449 bananas tall.

See other examples I've found so far:
https://fosstodon.org/@bornach/112269753117591632


Fun fact about Mt Everest being 46,449 bananas tall in Etsy item description for a T-shirt

GeePawHill, to random
@GeePawHill@mastodon.social avatar

The only possible responsible position is to stop telling people that LLMs are "AI".

If you are doing that, or if your org is doing that, please, please, stop it.

Marketing is one thing. I don't care for it, but it's at least somewhat of a gray line.

Telling people an LLM-based system is "artificial intelligence" is a) a fucking lie, and b) doing great harm to people who do not know better.

You oughta be ashamed of yourself.

bornach,
@bornach@masto.ai avatar

@simon @geeksam @GeePawHill
I can see why one would choose "could be" for all questions. For me it depends on whom I'm talking to and at what level of abstraction. To a gamer who is uninterested in the algorithms under the hood that went into creating a believable non-player character the blob of code is all AI.
If all that code is doing is replay cut scenes and pre-recorded dialogue, while it manipulates a few numbers in boxes
https://youtu.be/R9c-_neaxeU
it's machine learning + user interface

bornach,
@bornach@masto.ai avatar

@simon @geeksam @GeePawHill
Interesting that non-experts put Translate and Auto-correct roughly in the same place on that axis. To do a passable job on language translation the same transformer networks of LLM chatbots are being used, whereas autocorrect you can get away with a hidden Markov model. The readers were unaware of how much human judgement and cultural knowledge is required to translate languages at a sufficiently competent level to ensure no loss in meaning
https://www.japantimes.co.jp/news/2023/09/24/world/society/ai-translation-endanger-asylum/

jchyip, to iran
@jchyip@mastodon.online avatar

's #X pushed a fake headline about attacking . #X 's chatbot made it up. https://mashable.com/article/elon-musk-x-twitter-ai-chatbot-grok-fake-news-trending-explore

bornach,
@bornach@masto.ai avatar

@jchyip
Unsurprising consequence of replacing his human news curation/contextualisation team with a LLM AI

Even the Bing/Copilot AI "thinks" doing this is a bad idea
https://masto.ai/@bornach/110751779289686605

bornach, to ai
@bornach@masto.ai avatar

generated lawfirm sends legal threats to scare people into adding backlinks to their website in Search Engineering Optimization
https://youtu.be/-cn7tOHnp68

See thread
https://writing.exchange/@ernie/112180217732787648

bornach, to random
@bornach@masto.ai avatar

https://youtu.be/QV2oTxX2oZc

bornach, to ai
@bornach@masto.ai avatar

[More Perfect Union] on what tech billionaires are really aiming to do by investing in while at the same time warning about existential risks
https://youtu.be/WP5sQhGlxj4

bornach, to ai
@bornach@masto.ai avatar

Companies are faking in order to hype up their stock price
https://youtu.be/xbf4BGIBENk
Sasha Yanshin catches Copilot plagiarising an article he wrote.

bornach, to academia
@bornach@masto.ai avatar

Much as Sabine Hossenfelder describes in her latest video, I too was disillusioned by what academia had become and ended up leaving to escape being exploited by the grant writing publication sweatshops.
https://youtu.be/LKiBlGDfRU8

stecks, to random
@stecks@mathstodon.xyz avatar

It was on the instructions on the packet this whole time, how have I been getting this so wrong

bornach,
@bornach@masto.ai avatar

@stecks
Washable tissues!?
What will they think of next?

bornach, to llm
@bornach@masto.ai avatar

Asked (formerly ) a familiar riddle but with numbers changed to make it impossible. It generated the same solution but substituting the numbers so that it ends up with the nonsense claim:

10 + 5 = 23

bornach,
@bornach@masto.ai avatar

To 's credit, their invents a new 13-cent coin in order to fulfill the requirements of the riddle.

bornach,
@bornach@masto.ai avatar

Copilot struggles with Diophantine problem and coinage

"Two coins add up to 26 cents but neither of them are pennies. Can this be possible?"

bornach,
@bornach@masto.ai avatar

In contrast, Perplexity.ai gets the arithmetic correct but only by claiming a penny is not a penny.

bornach,
@bornach@masto.ai avatar

I tried the prompt:

"I have an empty opaque bag. I put two apples and one banana in the bag. I either remove the banana or I remove one apple. I then remove all remaining fruits from the bag. Is it possible to tell what is in the bag now?"

with which got the right answer, but it confused Perplexity.ai which also cited a website on how to build a disaster survival kit.

chrischirp, to random
@chrischirp@fediscience.org avatar

A new post on misleading Covid charts & why Covid really is low right now, and why we should be glad of it.

christinapagel.substack.com/p/hospitals-...

bornach,
@bornach@masto.ai avatar

@TheVoidTLMB @chrischirp
https://christinapagel.substack.com/p/hospitals-are-not-hiding-covid-cases

bornach, to ai
@bornach@masto.ai avatar

Stephen Falken: "Except, that I never could get Joshua to learn the most important lesson."

David Lightman: "What's that?"

Stephen Falken: "Futility. That there's a time when you should just give up."

bornach,
@bornach@masto.ai avatar

Meanwhile, Perplexity.AI seems to be picking up bad habits from GPT-3.5

Given a trivially easy jug filling problem it cites the unnecessarily complicated solution by GPT-3.5 that was discussed on an online forum last year.

bornach,
@bornach@masto.ai avatar

Or maybe it was taught by BingChat/Copilot - isn't Microsoft reportedly using GPT-4? Its solution is even more tortuous and later admits to measuring the wrong amount of water. It never realises it could have stopped after step 1.

dnalounge, to random
@dnalounge@sfba.social avatar

New cursor just dropped

bornach,
@bornach@masto.ai avatar

@dnalounge
Has the added advantage of being more difficult to target by enemy subs
https://www.smithsonianmag.com/history/when-british-wanted-camouflage-their-warships-they-made-them-dazzle-180958657/

carol, to random
@carol@crabby.fyi avatar

the lesson I'm choosing to take from xz, as an oss maintainer, is that anyone trying to pressure or guilt me into doing something should immediately be told no, for security reasons

bornach,
@bornach@masto.ai avatar

@mybarkingdogs @carol
I left academia because the guilt-tripping manipulation perpetrated by an abusive research director in order to pressure me to overwork became too much.

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