Hi
Hope the year of the 🐉 is off to an excellent start for you wherever you may be. Given everything back in Vietnam is shut for Tet, I am writing this in a wonderful cafe on top of the cliffs overlooking Aaliyirakkm Beach (Varkala, India) and loving life. With that, a few headlines from the world of AI for your Mondays:
Bill Gates has a podcast?!? Sam Altman on Unconfuse Me with Bill Gates
So who doesn’t have a podcast these days? Turns out even Bill Gates (whose $113 billion fortune makes him the 4th richest person in the world) has jumped on board the podcasting train. And one of his first guests? None other than OpenAI CEO Sam Altman.
They cover a lot of ground but Altman memorably describes GPT3.5 and GPT4 as “the stupidest these models will ever be”, saying that for at least the next 5-10 years we will be on a steep improvement curve. In practice, this will not be AIs that you will say “complete this email” or “write this code for me” to – it will instead be AIs to which you can say things like “Go start and run a company for me” or “Go discover new physics” – and it will do them. Bill Gates turns out to be a decent interviewer and notes concerns that he (and Altman) have to the counter to these positives – the fact that AI “will force us to adapt faster that we’ve had to … ever before”. Given Altman is who he is, his response is perhaps worth listening to in full.
IMF maps rocky road ahead for world labour markets
The IMF has released a major report analysing how AI could transform labour markets globally in the years ahead. The report reveals AI’s significant potential reach, with an estimated 40% of jobs globally exposed to AI transformation. However, exposure varies dramatically based on economic advancement – advanced economies may see 60% of occupations impacted as AI permeates even complex cognitive tasks, yet lower-income economies may encounter only 26% exposure. Further analysis shows a mixed outlook even for exposed roles. In advanced countries, nearly half of affected jobs are complementary, boosting productivity without worker displacement. But the remaining half see direct automation replacement, signaling major workforce disruption alongside economic gains.
Worryingly, individuals and countries already disadvantaged – women, youth, those lacking updated skills or digital infrastructure – appear poised to lose out most. The IMF urges swift, concentrated efforts on frameworks, retraining, and infrastructure investment so these groups and emerging economies can share in productivity and wage benefits rather than further divide access to AI’s possibilities.
Eager to upskill in AI but low on resources? We’ve got you covered
It’s been a minute since we’ve had a recap on this key question and when luminaries such as the Professor Learner Centred Design at the UCL Knowledge Lab are speaking to “an urgent need to equip teachers and policy leaders with practical skills and resources around AI”, we take that seriously. A quick breakdown of potential resources:
- Courses – Google’s Introduction to Generative AI Learning path is a good grounding to begin from before moving on to the Career Essentials in Generative AI by Microsoft and LinkedIn by which point you should have covered off the basics. After that, people might want to look into things like the deeplearning.ai offerings or those by various people on LinkedIn and other places (e.g., Ruben Hassid, Allie K Miller, etc.)
- Events – the 2024 AI in HE Symposium (Australia and NZ) was fantastic – keep your eyes out for recordings of that and other upcoming events from the organisers;
- AI in HE thought leaders – Ethan Mollick, Danny Liu, Conor Grennan, Jason Gulya, Amanda Bickerstaff – the list goes on;
- Videos – Practical AI for Instructors and Students is great as a starting point as is Two Minute Papers if you prefer to be out in the deep end;
- Books are starting to emerge – Ethan Mollick has one coming out soon, Leon Furze (who’s AI Assessment scale has received a lot of great press) has one available for pre-order, Rose Luckin has one too; and
- Newsletters – George Siemen’s Sensemaking, AI, and Learning, The Neuron (a general overview of the developing technologies), and AlphaSignal (fantastic but more technical).
All worth a look – but I’m not going to pretend to be the expert – what resources have we missed? Where else are people getting their AI info from?
AI celeb tutors – novelty engagement strategy or bridge too far?
Sometimes I wonder if we really think radically enough when it comes to leveraging AI capabilities for learning innovations. This one is getting out there though – how about a Kim Kardashian and Taylor Swift calculus tutor tag-team?
While AI-generated likenesses still occasionally trip into uncanny valley territory, they are improving rapidly (see, for example, this fairly recent post on HeyGen – hugely impressive and worrying by equal measure), the idea of harnessing pop culture icons that students love as virtual tutors is an interesting one. Is your child tuning out their math teacher but obsessively streaming BlackPink? No problem, get Lisa or Jisoo’s AI avatar to walk them through derivatives 24/7! The novelty factor alone could spur enthusiasm where traditional materials fail. And allowing learners to interact with mentors tailor-made to their interests unlocks engagement on a whole new level. Or is this just another fad that will push impressionable youth and fans into weird places? Who knows – but, either way, as these systems keep advancing, we’ve likely only scratched the surface of how AI-fueled personalised education could work.
Magnific.ai’s mind-blowing image enhancements: Long John Silver like you’ve never seen him before…
Magnific is a relatively new player in the AI image space but has made a big splash in very short order. Magnific’s groundbreaking image upscaling and enhancing tools have to be seen to be believed.
See the Magnific enhancement of an AI-generated picture of a pirate (before and then after Magnific upscaling and enhancement). And this works for “real-life photos” as well – check out the 128x zoom on a particularly excellent-looking piece of steak by Dogan Ural – absolutely mindblowing.
Long John Silver (Midjourney)
Long John Gold Class (above image x Magnific)