Professor of Ethics & Technology; AI & Natural Intelligence; Culture, Evolution, & Governance. See "website" for my social media policy. Was @j2blather @j2breve
The Estonian minister of justice complained about not seeing people like me debating AI policy here anymore.
We now again do that in the open. See the website link in my bio here for directions to where that happens.
X doesn't let me read DMs from a secure browser, so email me.
But just to say where some people are using the #AISE26#tag : I wrote THE paper on #AIBias back in 2017, and what matters most is when people intentionally do bad things with their AI. So what I want to work on is #AIRegulation.
https://t.co/QZDhm5JLJE
#AI governance: Rapid technological innovation continues to challenge existing policy frameworks, requiring int'l cooperation and cross-sector dialogue.
See how @UNIDIR’s Global Conference on AI, Security & Ethics serves as a catalyst for action: https://t.co/Sac6Kbcs9w #AISE26
@willknight ML being paid by GAFAM are still here for political reasons. A lot of AI people on bsky. I noticed @VHeddesheimer got the same number of RTs recently on twitter and bsky when he had a big announcement.
@willknight ha, sort of! 17K followers there, do you really not get more click throughs there than here? Or do think the musk machine lies about how many impressions our posts are getting?
Clive Lewis, said: “Democracies can change their mind - they’re sovereign.
You look at other countries that have stayed well away from Palantir – France & Germany. They’ve had the same briefings every other European country has, which is to allow a company like that, with the connections it has to unsavoury political views, is asking for trouble.”
@prmurphy But when I'm just asking a quick question, I do tend to use natural language including politeness, unless I find myself in a coding frame of mind, then I write code in NL.
PS https://t.co/y7z4dZWE10
@prmurphy eg–6. DON'T SHOW 'REASONING' JUST GIVE ME OUTPUT; correctness is easy to check. Don't waste energy.
8. These instructions are not asking for a new run unless otherwise specified, e.g. "do that again with the following rules." Otherwise, just type "acknowledged" & wait for input.
"AI overviews aren't search results
Search engine liability rules don't apply to AI "search"
Google's "users can check for themselves" defense falls flat
AI-generated opinions get less free speech protection"
Ha, reading the EU disclosure on blue ticks just reminds me how much X is NOT supporting the free speech I had in 2014, when thousands of people would have seen me say: THIS IS HUGE!
https://t.co/4wQtWSpIm1
#AIEthics
Argh! one reason I always use @Outlook in @googlechrome is so I can use emacs key bindings to move around my text quickly (like all my actually preferred editors.) But some consentless "upgrade" in the last 12 hours is making ^f do search!! Does anyone know how to turn that off??
So I ran the @apple@applesupport upgrade & now @google refuses to acknowledge my login. It doesn't say my password is wrong, it says it doesn't have enough info. I have always denied it a phone number, my home address etc. But apparently it "knows" I don't live in Austria.
@LinkedIn Relatedly, now @PayPal refuses to let me reset my password, "We’re having trouble making sure it’s you" That page has URL https://t.co/bXgOUJl4pE & no further options. I could make a new account with another email address, but why? (I NEVER use #paypal) What is it about #Austria?
Now @LinkedIn has a similar issue. It demands I confirm my ID with my phone, but then tries to create a new account for me, then complains about my existing account having the same email address! By its online support I can use my work email to confirm but there's no such option.
@j2bryson@Vivek4real_ Copilot can use a bunch of different models, with Claude being one of them. So that can help with costs, and Copilot may have ways, internally, to avoid using a bunch of tokens unnecessarily.
And sometimes, Microsoft likes to dog food their own products.
I use Copilot.
@eclipsed4utoo@Vivek4real_ That's what I thought – this sounds like the delta between Claude & Microsoft's own AI is smaller than the value Microsoft gets from forcing its programmers to improve its AI. https://t.co/ljTSiF8AX8