@mpesce@arvr.social
@mpesce@arvr.social avatar

mpesce

@mpesce@arvr.social

Futurist, VRML co-inventor, author, educator & speaker. Multi-award-winning columnist for The Register, columnist Cosmos Magazine, award-winning podcaster of both The Next Billion Seconds and This Week in Startups Australia. 🏳️‍🌈 Gadigal Land. (he/him)

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

jpm, to random
@jpm@aus.social avatar

Show me a phrase more cursed than “HR is thrilled to launch Workday for Microsoft Teams to all employees”.

I’ll wait.

mpesce,
@mpesce@arvr.social avatar

@jpm "IT is thrilled to launch Microsoft Recall to all employees."

mpesce,
@mpesce@arvr.social avatar

@jpm Do you reckon any of them know that?

mpesce, to random
@mpesce@arvr.social avatar

As someone who has been tracking this space closely for well over a year I can tell you that the 'unjustified hype' era of modern AI is well and truly over.

That doesn't mean there isn't going to be more hype - far from it. There's too much money riding on that for it to be otherwise.

But it does mean that everyone (and, in particular, journalists) now come at the field with a set of perspectives, critiques and questions that can quickly pierce holes in this most overinflated of balloons.

mpesce, to random
@mpesce@arvr.social avatar

These AI tools still hallucinate an alarming amount of the time: the Lexis+ AI and Ask Practical Law AI systems produced incorrect information more than 17% of the time, while Westlaw’s AI-Assisted Research hallucinated more than 34% of the the time.

http://windowscopilot.news/2024/06/04/ai-on-trial-legal-models-hallucinate-in-1-out-of-6-or-more-benchmarking-queries/

mpesce, to random
@mpesce@arvr.social avatar

Sh*t is going down, peeple, that I would not have believed could happen.

More...as it happens.

mpesce, to random
@mpesce@arvr.social avatar

Something Strange Happens to Wolves Infected by an Infamous Mind-Altering Parasite

"A study of 26 years' worth of wolf behavioral data, and an analysis of the blood of 229 wolves, has shown that infection with the parasite Toxoplasma gondii makes wolves 46 times more likely to become a pack leader.

The research shows that the effects of this parasite in the wild have been horrendously understudied – and its role in ecosystems and animal behavior underestimated."

https://www.sciencealert.com/something-strange-happens-to-wolves-infected-by-an-infamous-mind-altering-parasite

mpesce, to random
@mpesce@arvr.social avatar

Yesterday was the hottest day ever recorded in Delhi.

Until today.

And by 3 degrees.

https://www.theguardian.com/world/article/2024/may/29/delhi-temperature-hits-499c-as-indias-capital-records-hottest-day

mpesce,
@mpesce@arvr.social avatar

Breaking a heat record by three degrees - I am not a statistician, but that feels like it's four or five standard deviations away from anything like normal weather patterns.

mpesce, to random
@mpesce@arvr.social avatar

"Just because there exists a model that could read your organization’s entire Google Drive contents before answering each question doesn’t mean that’s a good idea."

The whole piece is gold.

https://www.oreilly.com/radar/what-we-learned-from-a-year-of-building-with-llms-part-i/

mpesce, to random
@mpesce@arvr.social avatar

"My boy, I have one word for you...

Plastics!"

Didn't quite work out as planned...

https://www.nytimes.com/2024/05/28/climate/pfas-forever-chemicals-industry-lawsuits.html

mpesce, to random
@mpesce@arvr.social avatar

A developer on Twitter has taken the discussion a step further by revealing that one of the flagship new features, Recall, doesn't require the presence of a powerful NPU after all.

http://windowscopilot.news/2024/05/28/windows-11-copilot-ai-features-like-recall-can-be-enabled-without-an-npu/

mpesce, to random
@mpesce@arvr.social avatar

Even at this early stage, though, Anthropic's research provides an exciting framework for making an LLM's "black box" results that much more interpretable and, potentially, controllable.

http://windowscopilot.news/2024/05/28/heres-whats-really-going-on-inside-an-llms-neural-network/

mpesce, to random
@mpesce@arvr.social avatar

This blog would never recommend using a prompt injection against a recruiter.

http://windowscopilot.news/2024/05/28/tip-how-to-work-with-recruiters-using-ai-to-filter-resumes-comedy/

mpesce, to random
@mpesce@arvr.social avatar

"Google and the other companies chasing this AI fantasy are desperate to have us see these systems as a life-changing leap forward, but it’s critically important for us to remain aware of the very real minuses that come with this latest shiny plus."

http://windowscopilot.news/2024/05/28/gemini-is-the-new-google/

mpesce, to random
@mpesce@arvr.social avatar

Another thing to consider: Sergey Brin came back from retirement to help oversee Google's AI work.

Which means a lot of this lands on him, and on Pichai, and on the board.

Who all seem to have no actual idea what they're doing.

An organisation worth two trillion dollars.

https://www.theverge.com/2023/8/18/23837372/command-line-google-co-founder-sergey-brin-ai

mpesce, to random
@mpesce@arvr.social avatar

FTC Chair Lina Khan said Wednesday that companies that train their artificial intelligence (A) models on data from news websites, artists’ creations or people’s personal information could be in violation of antitrust laws.

http://windowscopilot.news/2024/05/27/ftc-chair-ai-models-could-violate-antitrust-laws/

mpesce, to random
@mpesce@arvr.social avatar

It's actually pretty nice, showing only the traditional 10 blue links, giving you a clean (well, other than the ads), uncluttered results page that looks like it's from 2011.

http://windowscopilot.news/2024/05/27/google-searchs-udm14-trick-lets-you-kill-ai-search-for-good/

mpesce, to random
@mpesce@arvr.social avatar

"...I'd stumbled onto something bigger than a bug. Models from different providers use differing training datasets, machine learning algorithms, hardware, and so on. While they may all seem quite similar when dressed up with a chatbot front-end, each uniquely reflects the talents and resources used to create them. Finding something that affects all of them points away from the weakness of a single implementation, toward something more fundamental: A flaw..."

@theregister https://www.theregister.com/2024/05/23/ai_untested_unstable/

mpesce, to random
@mpesce@arvr.social avatar

Just submitted my topics for this week's RNZ NineToNoon segment (Thursday 9am AEST) and dear god, you could not make these up: The 'LLM Kryptonite' bug I found and tried to report; Microsoft's 'Totalled Recall', and Google telling everyone to eat rocks.

Charlie Brooker, thou art avenged!

mpesce, to random
@mpesce@arvr.social avatar

this whole AI-in-the-search-results thingy feels like such an unforced error one wonders how it made it through all the processes intended to protect their crown jewels?

perhaps those processes don't exist? or were ignored in panic?

https://www.nbcnews.com/tech/tech-news/google-ai-im-feeling-depressed-cheese-not-sticking-to-pizza-error-rcna153301

mpesce, to random
@mpesce@arvr.social avatar

in other news, it certainly seems clear that Google has made the decision to pivot from search to light entertainment

mpesce, to random
@mpesce@arvr.social avatar

"Social media companies that conduct business in a jurisdiction are subject to the laws of that jurisdiction."

That's the law, at least in Queensland, and quite possibly all of Australia, given what I've just heard on ABC News.

mpesce, to random
@mpesce@arvr.social avatar

Now it can be told:

While doing some AI engineering work for a client, I developed a prompt - completely inadvertently - that reduced every AI chatbot to gibberish (except Anthropic's Claude 3). I then spent a week trying to alert the LLM vendors to this issue - and largely failed. There is no mechanism to report flaws in these models that are already deployed to billions of users. Read the whole story in @theregister

https://www.theregister.com/2024/05/23/ai_untested_unstable/

mpesce,
@mpesce@arvr.social avatar

@researchbuzz @theregister I am not entirely sure the security team has been briefed on prompt injection attacks but i could be wrong about this

mpesce, to random
@mpesce@arvr.social avatar

A month ago I accidentally coded up a prompt that seems to break all of the AI chatbots, except Anthropic's Claude. I tried to report it to the various vendors - only to learn there's no mechanism to report these kinds of flaws...

http://windowscopilot.news/2024/05/24/i-stumbled-upon-llm-kryptonite-and-no-one-wants-to-fix-this-model-breaking-bug/

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