Hi
Hope you had excellent weekends. Mouldy March is holding on a bit longer than we’d like in Hanoi but, balanced against that, it’s a short week with the Hung Kings Festival on Thursday #winning. A few headlines from the world of AI to kick off the short week:
Is 2024 the Year of AI agents?
So about 100 years ago in AI time, in April 2023, an interesting paper was released that explored a simulation where 25 AI agents were given motivations and memory and simulated a small town together. The results were crazy – the agents engaged in daily routines, formed relationships, discussed the upcoming mayoral election, and threw a Valentine’s Day party for the whole town – paper here and overview from the wonderfully effervescent Károly Zsolnai-Fehér below. What a time to be alive!
Anyway, things have gone a long way since then – we’ve mentioned Devin in these updates before – but that’s just the beginning. Andrew Ng has been talking extensively about multi-agent workflows being the next big thing (great outline of multi-agent workflows below) and there are papers coming out that suggest that it’s not clever prompting that really makes the difference – it’s having AI agents collaborating on tasks, an effect that seems to scale with increased difficulty. Super interesting stuff…. AI agents are definitely a space to watch this year.
The giant is slowly waking… Google Unveils Generative AI Innovations at Cloud Next ’24
OpenAI and Microsoft clearly got the jump on Google late 2022/early 2023 – and Google definitely hasn’t helped itself with some truly historic dropped balls like Bard’s initial launch (a $144bn marketing mistake) and the issues with Google Gemini’s image generator (a $90bn “miss of the mark“). That said, Google invented a wide range of the technology underlying this revolution and, once they figure out how to leverage their enormous data reserves, well… things get very interesting very quickly. After all, combine YouTube, AdSearch, Search history and you already have enormous libraries of content and personalised knowledge – and that’s before you remember that 70% of the world’s smartphones run on Android.
Google’s Google Cloud Next ’24 event unveiled groundbreaking generative AI advancements, including Vertex AI Agent Builder, which allows businesses to create powerful customer service agents using the Gemini 1.5 Pro language model, and seamless integrations of generative AI within the Google Workspace suite, such as Gemini for Google Workspace and Google Vid, an AI-powered video creation tool, empowering enterprises to enhance customer experiences, boost productivity, and drive innovation. Keynote supercut below.
AI’s Rapid Ascent: Superhuman Performance and Autonomous Agents on the Horizon
In a timely update from Ethan Mollick (whose excellent new book Co-Intelligence: Living and Working with AI is very much worth a look by the way), What just happened, what is happening next explores the rapidly evolving world of A as, the capabilities of large language models continue to expand at an astonishing pace, with the current frontier models like GPT-4, Claude 3, and Google’s Gemini demonstrating “superhuman performance” in various human tasks such as persuasion, medical reasoning, and specialised knowledge tests.
As discussed above, the development of autonomous AI agents has also progressed faster than anticipated, potentially leading to significant changes in how organisations integrate AI into their workflows. As the next generation of AI models emerges, the combination of their unclear upper capabilities, “superhuman” abilities, and the rise of autonomous agents could have profound implications across various fields, making it crucial to closely monitor the development and consequences of these advanced AI systems. Topically, this is possibly a good moment to review the AI Dilemma by Tristan Harris and Aza Raskin (below) or their follow-up JRE podcast – concerns of the nature outlined in these talks are becoming increasingly present and real.
AI Transparency Takes Centre Stage: New Legislation Aims to Protect Creators
The introduction of the Generative AI Copyright Disclosure Act in the US Congress and the European Union’s AI Act highlights the growing concern over the use of copyrighted material by AI companies. As generative AI technologies continue to advance, lawmakers and creative industry stakeholders are pushing for increased transparency and accountability to protect the rights of creators and their IP.
The Generative AI Copyright Disclosure Act, supported by various entertainment industry organisations, would require AI companies to disclose the copyrighted works used in their training datasets before releasing new generative AI systems, while the AI Act recognises the challenges and opportunities posed by general-purpose AI models to the creative landscape. As these legislative efforts progress, striking a balance between fostering innovation and protecting creators’ rights will be crucial in shaping the future of AI and its role in the creative world. #interestingtimes 👀
Something for the Creative Space
It’s been a minute since we’ve looked at what’s happening in this area. That’s not for lack of action, or interest, it’s just there’s been so much on…. that said, two incredible new developments: Udio and Viggle.
Udio (strapline: make your music) is an AI music generator that goes a lot further than some of the muzak generators we’ve had as yet – free to use and with the ability to add tags to refine your prompting, it gets interesting. Case in point, check out my weekend offering, prompt: taking my dog for a walk down the park Male vocalist, international afrobeat, Soul, Mellow, Synth funk, Psychedelic – it gave me “Leashed Dreams” (below). Admittedly it’s no Fela but a big step forward and pointing to some pretty interesting possibilities ahead.
Or Viggle (strapline: controllable video generation) – with this, you can generate images and then animate them according to i) prompts or ii) a source video. Runway is obviously great but there results here are pretty spectacular – case in point, apparently Joaquin Phoenix’s Joker is hyping his upcoming movie by doing festivals.
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. 🙏