Hi
Normal service resumes… hope all’s well with you and yours wherever you are. A few notes from the world of AI to kickstart your week:
AI Arms Race: Big Tech’s Latest Moves
There have been a lot of big technical announcements either released or teased in the last week or so:
- Anthropic (the people behind Claude – my own personal go-to writing companion) have launched a Team plan and an iOS app. Access to all of the Claude 3 models, higher usage, that incredible 200k context window and – interestingly – access to their upcoming collaboration features, data respositories, etc. It’s not ChatGPT Plus yet but this is not a Pepsi vs Coke comparison – there are definite upsides to Claude already so at $30/m maybe worth a look if you have the $ to spend.
- OpenAI are reportedly shaping up to take on Google directly – with an internet search engine tipped for release soon. Overview here and initial reporting back in Feb from the Information here. This is an interesting move and OpenAI are not alone in this – cool kids the Browser Company have a browser called Arc that effectively creates a webpage for individual queries using AI. An interesting alternative and potentially worth a look.
- Sam Altman (OpenAI) is talking big about GPT5, describing GPT4 as “the dumbest model any of you will ever have to use”. In an interesting talk on “The Possibilities of AI” at Stanford, he spoke at length about the opportunities and risks of AI and AGI – noting some big investments coming there in the future.
The Rise of Digital Persons and the Deepfake Dilemma
It’s been fascinating watching this whole space evolve as people work through the implications of how these emerging AI technologies can be used. We’ve mentioned Channel 1 – the AI News Channel before – wild stuff and I see now that the offer has evolved to be a more “personalised” version of global news – wonderful stuff on paper but what will that do to things like the Overton Window and broader ideas of political discourse?
You think it’s bad now? The potential for deepening schisms is concerning…
Or how about Ukraine unveiling the world’s first AI-generated foreign ministry spokesperson – a “digital person” named Victoria Shi? Good for messaging consistency if nothing else, I guess – see her in action below…
flipside (?) to this, did you hear about the US principal framed as racist when a disgruntled teacher made a deepfake recording of their principal saying various racist and antisemitic comments?
The field is moving extraordinarily fast and, while there are no doubt benefits to be realised, the potential implications and applications here are concerning. Topically, I see Dr Mike Perkins is doing a session on deepfakes as it appends to assessments on this very thing for ASCILITE next month. Might be worth a look.
From Simulations to Reality: 🐕 + yoga ball = 🤯
Nvidia do more than “just” chips. Not satisfied with selling spades in a goldrush (I see you Blackwell GPUs), Nvidia’s scientists are leading the charge with simulation —> real-world training. You might have heard of the Eureka AI agent that was training robot hands to spin pens in a simulation late last year? Welp, that has led to extraordinary scenes like this – a robot dog balancing and walking on a yoga ball IN REAL LIFE – no tweaking, no fine-tuning, it just works. 🤯
I once made the mistake of asking Claude how much faster it thinks than me using metaphors from the natural world – I won’t tell you what it said but it was uniquely humbling/I’d encourage you to ask an AI that yourself. Either way though, think about the implications of running simulations for learning – at computer speed – and then uploading that knowledge into a robot in real life. Makes Neo in the Matrix’s “I know kung fu” line look cute…
“Still hiring humans?” Bland AI – and what we can do about this
A great piece of marketing, a billboard in the US had this excellent … provocation? Call to action? Whatever it was, it gave a phone number to an AI doing customer support and receptionist work. Scale that up to the millions of people there must be working in call centres and we have a major disruption here. Expand it out to other fields… you can see where this might go.
So what can we do about it? Conor Grennan (NYU Stern School of Business) is an interesting listen in this space with good news for generalists who are, he says, “in an outstanding position to crush generative AI in your workplace. Here’s how…”. Very much worth a listen. Oh, and if you want to try Bland.ai itself, you can do that here.
AI comes to Springfield…
Don’t normally love me a reboot but I have to say this 1950s version of the Simpsons is awesome. All the characters are instantly recognisable – though Ned Flanders has more Bryan Cranston than I’d’ve gone for and Marge has a bit of a haunted look to her eyes. Still – worth a look! “I am the one who knockily-dockilies…” 👀 🍿
Marge Simpson (nee Bouvier)
Homer J
We hope this edition of the newsletter has been of interest to you. If you’re new here and it’s been useful, please do click subscribe and you can expect a weekly update every Monday from now on. If you’re already a subscriber – thanks for your ongoing interest and support! Either way, if you know others who might benefit from reading this weekly, please forward it on to them too.
Have a great weekend ahead and let us know if there’s anything we’re missing that we should add to make this newsletter more useful for i) yourself and/or ii) others. This is a fast-moving, ever-evolving space and we greatly value any and all feedback.