Happy Monday – I hope the weekend treated you well and the week ahead is shaping well. A few notes from the world of AI to start your week off right:
Putting humans first – a rational, humanistic “Third Way”
Following on the steps of Europe’s Artificial Intelligence Act (2023), the US Securities and Exchange Commission has rejected requests from Apple and Disney to exclude conversations about their usage of AI from annual meetings calls. Interesting legal reflections are emerging, suggesting this could be the beginning of “shareholder activism when it comes to AI deployment”. AI cheerleader that I am, I’m encouraged to see early potential steps towards what UNESCO describe as “a human-centred approach to AI” (UNESCO, 2023: p18).
Jailbreaking AIs + human psychology = a cautionary tale
AI are weird. For every piece of evidence that makes them seem alien, there is often another that seems almost too human to be true. For example, a paper demonstrating just how easy it is to overcome even advanced AI guardrails by using persuasion techniques that work on humans has just been released. Drawing deeply from the (human) psychology literature, How Johnny Can Persuade LLMs to Jailbreak Them is an illuminating read – for anyone working with student-facing GPTs or, in fact, anyone with important IP tucked away in a GPT on the new OpenAI GPT store. On that, reports are coming in that copycats are everywhere – with even OpenAI’s own GPT’s being copied!
Household Help to Handhelds: Emerging AI assistants
Does anyone remember Rosey the robot from the Jetsons (see gif at bottom)? Seems like students at Stanford do – and they’ve just made the first version. Called “Mobile ALOHA”, the robot performs various household chores by observing and mimicking human actions. Much cheaper than previous robot assistants, able to be controlled remotely (so there’s pretty wild implications there), and with the ability to learn from a small number of human demonstrations, this is great news for people who don’t like to cook or clean (and, I’m sure, a whole lot more in time). Demo and github here, a breakdown here, and a behind-the-scene bloopers/fail reel here.
Rosey the Robot (The Jetsons)
The Rabbit r1 though, feels like a brand new class of personal device – big talk for something this orange (see pic) but it might just be the smartphone killer. Built on a new form of hardware and with a brand new OS, the r1 “removes” apps from the user experience. From simple tasks like booking an Uber or a pizza to complex ones – like planning detailed itineraries for overseas travel and booking the tickets for you – it does the lot. And all this just by talking with it! With a response speed reportedly 10x faster than ChatGPT, this is already very impressive but they say this is just the beginning. Watch the keynote here!
The Rabbit r1
A timely reminder on AI, Turnitin, and Exams
This is perhaps worth repeating as the new year starts: AI-generated text cannot reliably be detected. Full-stop. And if anyone tells you otherwise, maybe send them this slightly dated now, but nonetheless thorough and very human evisceration of the argument by Sarah Eaton (University of Calgary). The follow-up might be: “Right, well let’s go back to exams” (not at RMIT of course but still…) – the Conversation has an excellent review of the arguments for and against their use in HE. Spoiler: there’s “surprisingly little hard evidence to justify their their widespread use in university assessment”. Authentic assessment ftw!
So there are AI students now? 🤖 🧑🎓
Ann and Fry are two freshmen at Ferris State University in Michigan. They’re part of a hybrid delivery model where they will work with classmates, complete assignments, etc. The twist? Ann and Fry are virtual students powered by AI. Apparently the idea is to “help faculty members find new ways to make education more accessible, among other uses on campus and in the community”. Reportedly being built in combination with the US Dept of Defence, NSA, and Homeland Security, it’s a different kind of experiment. Esp. when you consider that, awesomely, Ann and Fry have their own fake back stories which will be used to help them select their own majors – wonder what they’ll choose?
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 an excellent week and catch you next time!