Steve Jones

A2: zero-trust now needs to extend to human identity. When responding to AI threats you can only succeed if your AI is ahead, so a key additional question is how you manage the human scope of risk, knowing that AI can act like a user.

A1: TL;DR: Panic

Longer answer: if you’re not including AI based social engineering industrialisation into your cybersecurity risk focus and adaptive malware enabled by AI then it’s liable to be an « exciting » cybersecurity year.

Sad to hear of Tony Slattery’s passing, even sadder that he’s a case study in AI misinformation

According to Google he’s the son of the senior editor of the Atlantic, who appears to be younger than him.

Welcome to no fact checking.

The new #AppleSports app was clearly developed by people who don’t know sports outside of the U.S.:

No domestic football cup competitions (so apparently Wolves had no game today)
No IPL or indeed any cricket
No Rugby
No internationals

The claims of AGI being a “solved problem” make me think about the fact that Climate Change is effectively a “solved problem”, but people are claiming we need AGI to solve it, rather than actually doing what needs to be done.

2025 will see many claims of AGI

The best carry on has two wheels, four wheels reduces the packing volume.

The rise of AI fake news is creating a ‘misinformation superspreader’ Artificial intelligence is automating the creation of fake news, spurring an explosion of websites that can disseminate false information about wars and elections

Do you have evidence that it hasn’t had disastrous effects?

www.washingtonpost.com/technology/2...

🆕 Out now: FLI's 2024 AI safety scorecard! 🦺

🧑‍🔬 We convened an independent panel of leading AI experts to evaluate the safety practices of six prominent AI companies: OpenAI, Anthropic, Meta, Google DeepMind, xAI, and Zhipu AI.

🧵 Here's how it went 👇

Just a reminder: Barbers and Hairdressers in most places are subject to more regulations than AI companies

Just a reminder that Pharma and banking are two of the most regulated industries, the first can still cure diseases, and the latter still managed to invent CDOs

Regulation does not mean no innovation.

AI regulation can absolutely drive accountability without stopping innovation.

2025 Prediction: Shadow IT will become LOB AI Why career success will dictate who runs AI

#2025Prediction the business buying AI Agents directly will result in LOB AI being the main driver of #AI.

blog.metamirror.io/2025-predict...

You can be suspicious though if they’re claiming it for commercial and market valuation reasons.

First day in the bag for the season, November boarding at Sunrise Ski Resort, White Mountains Apache Tribe owned and operated @lakotaman.bsky.social

Absolutely, the debate prior to 2022 was about bias in AI, which doesn’t feel terribly right wing.

And with all of this we will continue to find even less images of the Loch Ness Monster or Bigfoot.

An high end phone today is better than most SLRs from the 90s, and the top full frame and medium format cameras of today can take high iso images that would have been impossible.

The sensor quality is unlikely to get 10x better in a decade, but storage will continue to increase as will sensor response time.

This basically means we’ll be taking 50MP 120fps “photos” at some stage and then AI will help us pick, or create the best picture from a set.

However two decades ago with film I would take a few hundred shots (ten rolls of film) that would cost a fortune to process and I’d get less good photos because you’d choose your shot.

A decade ago I’d have a few gig and I’d need spare cards for a holiday. Now I can leave stuff on there for months before I need to clean up.

But a decade ago I’d take thousands of shots to get a few good ones.

Just bought a 1TB SD card to go with the CF Type B 1TB card in my camera.

This basically means I’ll have the camera on “take a bazillion photos” which underlines the evolution in photography.

Debugging Agentic code is like debugging code written by a printer. It has all it requires to work, but it chooses to pretend it doesn’t.

Well found @molly.wiki and @web3isgoinggreat.com so that’s the crypto-popcorn entertainment moved over to Bluesky.

Agentic Integration This isn’t just an API problem

This and AgentProtocol from LangChain are opening up the whole new field of Agentic implementation, and it won’t be trivial.

blog.metamirror.io/agentic-inte...

NIMBY - Not in my back yard
NOTE - Not over there either
BANANA - build absolutely nothing anywhere near anything

The housing crisis is BANANA created.

Trigger warning

Where is #AI Bluesky? Lots of politics but I’m not seeing much tech

In a culture where sometimes it feels like people value speed and noise over experience, take a day, be like Ferris.

“Life moves pretty fast. If you don't stop and look around once in a while, you could miss it.”

When I first moved to the U.S. the standard carton milk weirded me out, multi-month shelf life and terrible taste. Then I discovered local dairies where the milk has a shorter shelf life and tastes better.

Both are pasteurised.

At no stage did I think raw milk was the answer, because it isn’t 1861

Flying into Washington DC, tomorrow to talk about AI and government with Informatica and Databricks.

Could be interesting

Exactly, that is part of the challenge, how do you hand two departments collaborating let alone multiple different businesses.

It’s going to be interesting finding out.

Agentic Integration This isn’t just an API problem

#Agentic Integration, an interesting new challenge for 2025, and no, it isn’t just an API. #AI agents add a whole bunch of complex nuance to integration.

blog.metamirror.io/agentic-inte...