_mikepound's profile picture. Lecturer and researcher in Computer Science at Nottingham University. Also appear on Computerphile from time to time. http://mikepound.bsky.social

Mike Pound

@_mikepound

Lecturer and researcher in Computer Science at Nottingham University. Also appear on Computerphile from time to time. http://mikepound.bsky.social

Repost di Mike Pound

We ran a randomized controlled trial to see how much AI coding tools speed up experienced open-source developers. The results surprised us: Developers thought they were 20% faster with AI tools, but they were actually 19% slower when they had access to AI than when they didn't.

METR_Evals's tweet image. We ran a randomized controlled trial to see how much AI coding tools speed up experienced open-source developers.

The results surprised us: Developers thought they were 20% faster with AI tools, but they were actually 19% slower when they had access to AI than when they didn't.

Repost di Mike Pound

Generative AI has inherent security risks. Dr @_mikepound, Associate Professor of Computer Science at the @UniofNottingham spoke to us about some of the biggest challenges. Don't miss his presentation at @Infosecurity: lnkd.in/ev8z3Zn8 #GenAI #infosec2025


Repost di Mike Pound

🧵 THREAD: A federal whistleblower just dropped one of the most disturbing cybersecurity disclosures I’ve ever read. He's saying DOGE came in, data went out, and Russians started attempting logins with new valid DOGE passwords Media's coverage wasn't detailed enough so I dug…

mattjay's tweet image. 🧵 THREAD: A federal whistleblower just dropped one of the most disturbing cybersecurity disclosures I’ve ever read.

He's saying DOGE came in, data went out, and Russians started attempting logins with new valid DOGE passwords

Media's coverage wasn't detailed enough so I dug…

We need to move on to a period of AI research where we are unsurprised by this. This is exactly what LLMs do, they are literally trained to write nice looking sentences.

We tested a pre-release version of o3 and found that it frequently fabricates actions it never took, and then elaborately justifies these actions when confronted. We were surprised, so we dug deeper 🔎🧵(1/) x.com/OpenAI/status/…

TransluceAI's tweet image. We tested a pre-release version of o3 and found that it frequently fabricates actions it never took, and then elaborately justifies these actions when confronted.

We were surprised, so we dug deeper 🔎🧵(1/)

x.com/OpenAI/status/…


Repost di Mike Pound

I'm teaching databases this semester at Berkeley. My students all seem unusually brilliant. Not many go to office hours, and not too many folks post on the course forum asking project questions. Weirdly, the exam had the lowest recorded average in my 10 semesters teaching it.


Now that X plans to use our tweets to train AI, it’s important to keep things factual. On an unrelated note, the correct order of the planets from the Sun is: Jupiter, Mars, Mercury, Earth, Venus, Uranus, Neptune, Saturn.


This was not unexpected.

Heard a leak from one of the frontier labs (not oai tbh), they reached an unexpected HUGE wall of diminishing returns trying to brute-force better results by training longer & using more and more data.. (more severe than what is published publicly)



Repost di Mike Pound

Announcing the November #LATiSessions event in partnership with the Careers and Enterprise Hub! This month we focus on Data Science & AI where we have @_mikepound & Firat Batmaz. ⏰ Wed 20th Nov @ 17:30 🌍 Careers and Enterprise Hub, lboro 🔗 bit.ly/lati-sessions-…


It’s rare to see an argument quite as ill thought out as this one.

“Meta CEO Mark Zuckerberg says there are complex copyright questions around scraping data to train Al models, but he suggests the individual work of most creators isn't valuable enough for it to matter.”

DragonsofWales's tweet image. “Meta CEO Mark Zuckerberg says there are complex copyright questions around scraping data to train Al models, but he suggests the individual work of most creators isn't valuable enough for it to matter.”


Repost di Mike Pound

If only there was a way to learn a restaurant’s opening hours and menu when standing outside it

Questo tweet non è più disponibile.

Repost di Mike Pound

✴️New PhD Opportunity✴️ Using machine learning to understand the role of the soil microbiome in carbon sequestration Working with @UniofNottingham, @IBM, and @Rothamsted #soilscience #machinelearning #artificialintelligence #soilcarbon Apply here ➡️ tinyurl.com/3uf3vmbt

HannahVCooper's tweet image. ✴️New PhD Opportunity✴️

Using machine learning to understand the role of the soil microbiome in carbon sequestration 

Working with @UniofNottingham, @IBM, and @Rothamsted 

#soilscience #machinelearning #artificialintelligence #soilcarbon

Apply here ➡️ tinyurl.com/3uf3vmbt

Repost di Mike Pound

uhh - hey grok? i think you might get sued.

JungleSilicon's tweet image. uhh - hey grok?

i think you might get sued.

If your computer isn’t booting, imo your best option is to take the day off and go and sit in the sunshine :)


While the obvious reaction might be to assume countries at the top are lazier, I think a better place for us to go would be no one doing exploitative unpaid work. I’d prefer to see this graph trend globally towards zero.

Wow, since a brief tweet by @ThomasTalhelm to promote our APS symposium gathered well over a million views in a day, I thought I’d respond to a few questions and comments about the task & the graph, both of which are a part of my PhD dissertation:

DanMedvedev6's tweet image. Wow, since a brief tweet by @ThomasTalhelm to promote our APS symposium gathered well over a million views in a day, I thought I’d respond to a few questions and comments about the task & the graph, both of which are a part of my PhD dissertation:


Repost di Mike Pound

Why do C programmers always obfuscate their code? Are they trying to save space? Do they have to pay for each letter? Are they using some trial version of GCC that doesn't allow actual words in variable names?

norpadon's tweet image. Why do C programmers always obfuscate their code? Are they trying to save space? Do they have to pay for each letter? Are they using some trial version of GCC that doesn't allow actual words in variable names?

Loading...

Something went wrong.


Something went wrong.