Michael Conway

@rorysutherland @daveg You could sell tickets - I’d be there!

@Michaelconway87 @daveg My daughter did this. Agree. Pain in the arse to argue with, though.

@daveg @rorysutherland Ha! As a practitioner running these types of teams, I genuinely think this answer is you need a broad church. Too much of any bias leads to sub-optimal outcomes. The variety is where the magic is. Long time listener, first time caller - thanks for the chat 😊

Yes, but like you say, it depends how it's taught e.g. Propositional Calculus is the foundation of computing. I'm an architect by background and that is sometimes an arts and sometimes a science degree. Social sciences should by rights be very important, but today's social scientists shouldn't be let near anything serious, with a barge pole.

@daveg @rorysutherland I raise you a philosophy degree in particular (I’m deep conflicted by holding one ☺️). Typically seen as a liberal arts degree, but Analytic philosophy (taught in all good Anglo-American unis as the defacto) is the most fiercely logical and reason based degree there is

@rorysutherland Hiring people that are less trained in reason against a machine that sometimes hallucinates seems like the opposite of what they should be doing.

@emollick @kevinrose Jean Claude Van Damned

Lots of terms floating around for "let the AI tell me what to do" and I just follow its instructions: being ratatouille'ed (@kevinrose), reverse centaur (Doctrow), Marcus Chen-ing (my latest proposal, as this is the generic name for a human used by AI most). Any other examples?

@tomfgoodwin Chatbots are twofold. Understand and Fulfil. LLMs solve understand but nothing has yet properly cracked the fulfil as it’s an integration problem and not an AI one.

A lot of technical people don’t real realize that a good chatbot isn’t where the entity feels like a human. It’s when it can solve a problem that you have, it’s about read and write access and permissions and systems and databases that talk to each other. All things that don’t seem to have improved.

@SpursOfficial 💙💙💙💙

Given today’s horrific events in Sydney, we recognise wishing our Jewish fans a happy Hanukkah may seem hollow at this incredibly difficult moment. The story of Hanukkah reminds us that even in darkness, we can find the light. We stand alongside our Jewish communities and pray for peace and unity across the world, while sending our deepest condolences to those who have lost loved ones 🤍🕎

@NottsPhilosophy #solipsism

Typo. We have more than one student #pluralism

Week one! Excited to start philosophising with our student 😀😀 #philosophy #teaching #uon #ready

@tomfgoodwin @emollick has a wonderful term for that exact phenomenon, he calls it the “jagged frontier”

I continue to be amazed at how mesmerizing good AI can be at some things, and mystified as to how bad it can be at other things. And even knowing a bit about how it works, it's generally baffling to me as to why, and I sense to many people who know far more.

@ItchyGin @nicolelampert It reads “chai” in Hebrew and literally means “life”. Very strong symbolism in Judaism of this word, as well as the number 18 it also symbolises which is a really lucky number as well.

@nicolelampert I've not seen this symbol before. What does it mean?

As the ceasefire officially ends tonight, actor Jason Isaacs wears the yellow ribbon as he hands out an award at the #britawards to remind people about the hostages who have now been hidden in tunnels for more than 500 days.

@BristOliver Schrödinger's CatGPT

I wouldn't claim to understand all the words in here nature.com/articles/d4158… but it seems like there's a superposition of physicists who simultaneously do and don't believe in it

Getting ads for (alleged) quantum chips on here now, and TBF that is a step up from the usual dropshipping and crypto

@emollick If only there was a readily available tool for summarising long docs 🤔

I find it odd to hear comments that the results of OpenAI Deep Research are too long, I think that misses the point of an analyst report You want to see a discussion of the evidence & argumentation. Reports are less for answers than to inform your thought in ambiguous situations

@emollick so we’ve focussed on quality & performant models which have transparency of where the data came from and OS’d via Apache 2. Then we have a bunch of value add services like watsonx to tune, train & govern your models, as well as consulting services to ensure you get biz value 2/2

@emollick Very much have a horse in the race here (I work there) but IBM has taken the Red Hat approach (as I guess you’d expect). It was proven pretty early on that there wouldn’t be a moat in the models (1/2)

I wonder if worrying about beating out other players in open AI models only really makes sense if you think that the end goal is to produce a closed AGI model that gives you advantages over everyone else. Otherwise what is the value of investing in your own model vs using one?

@francesweetman She’ll go into the holding pen with Esther Rantzen

@tomfgoodwin @emollick Or intelligence

@emollick Or AI

With no common definitions of agent, reasoning, copilot, tool use or autonomous, the next year in AI is going to be a lot of people talking past each other!

@tomfgoodwin An agent who does all the work for me, knowing my requirements and dealing with all paperwork both at work and at home is pretty transformative to my mind. Can imagine a world where websites disappear as Agents are booking in a backend api. Don’t think that’s a patch.

@Michaelconway87 So crawling and scraping existing sites. Which is my point. A workaround / patch

AI Agents always seem like Fingerbots. Am I missing something? Patches on systems vs AI first systems ?

@tomfgoodwin Talking in hypothetical capability, rather than current viability obviously…

@tomfgoodwin A “Book me a holiday” agent would search for best price/quality blend of options that it’s learnt from you and present you them. Give you a couple of top options and you tell it which one you’d like to book and does. Then will carry out all your HR stuff timeaway stuff at work

@anothercohen watsonx