UN, Tech Giants, and Cities Embrace AI’s Promise and Perils
Savor the radical Briefing you adore, in a format perfected for clicking and wine-sipping indulgence.
The Thin Wisps of Tomorrow
UN Unites for Safe, Beneficial AI. In a historic move, the UN General Assembly has unanimously adopted the first-ever resolution on artificial intelligence, recognizing the urgent need for global consensus on AI governance to ensure its safe, secure, and trustworthy development. The resolution, which aims to close the digital divide and leverage AI to achieve sustainable development goals, reflects a growing trend of countries such as Italy, France, Germany, and the United States to regulate AI. It is remarkable that 193 UN member countries can agree on regulating AI, but seemingly not much else.
AI Agents Are Iterating Towards Greatness. When Andrew Ng speaks about AI, we listen. In a recent letter, he emphasized that the future of AI lies in developing intelligent, multistep agents capable of iterating over tasks, reflecting on their work, utilizing tools, planning ahead, and collaborating with other agents. This argument aligns with what we’ve been suggesting for some time – recent demonstrations have shown that by incorporating these iterative agent workflows, even older language models can significantly outperform the latest models operating in the currently prevalent zero-shot mode. GPT-3.5 outperformed its much more capable sibling, GPT-4, by 40% when using an iterative agent workflow. As open-source agent tools and academic literature proliferate, it’s an exciting time to explore the potential of these design patterns to drive massive AI progress. Fasten your seatbelts, it might get wild out there.
The AIs Are Watching You. San Jose, Silicon Valley’s oft-overlooked sibling, contemplates deploying AI-powered cameras to identify homeless encampments. The NYPD, Gotham’s guardians, investigates AI that can spot firearms drawn in the subway. Meanwhile, MIT reveals that household robots lack the common sense to offer much utility. But take heart—although algorithms may soon surveil our streets and subways, our homes remain mercifully free from the prying gaze of android eyes (if only because they’re still too dumb to judge).
What We Are Reading
🔄 How to Actually Execute Change at a Company Unsurprisingly, only 60% of the planned value is typically realized in change initiatives. Learn what needs to be in place to turn the potential of an idea into realized value. Hint: Execution! @Jane
🚀 Why Some Companies Grow amid Uncertainty — And Others Don’t Strategic thinking and timely decision-making are two critical factors that help companies thrive amidst uncertainty. @Mafe
🍔 The Revolution That Died on Its Way to Dinner If you periodically find yourself wondering what happened to your lab-grown burger, here it is—on the other end of the hype cycle where a fantastically exciting, literally lab-grown proof-of-concept meets the challenging reality of engineering at scale. @Jeffrey
🤯 Among the A.I. Doomsayers The idea that oil companies and climate activists are part of the same community is a striking comparison. It introduces the chaotic system and network of people discussing doom and self-proclaimed effective accelerationists, making for a fascinating read! @Julian
🤑 The Curious Case of Thomas Welton Stanford The funds from Thomas Welton Stanford’s peculiar obsession propelled Stanford University to the forefront of modern psychological research. @Pedro
🌀 The Allergy to Uncertainty Embracing the beauty of uncertainty—discover how letting go of certainty can lead to a more fulfilling and curious life. @Pascal
Bits & Pixels
» Put your hands up in the air! Brace yourself for a plethora of fresh music powered by AI—SoundryAI’s innovative sample generator makes crafting your next house track a walk in the park.
» Chain-of-thought reasoning can enable neural networks, such as large language models, to tackle intricate problems that they would otherwise grapple with—computational complexity theory may help forge innovative approaches to problem-solving with AI.
» An adhesive neck patch, harnessing the power of AI, is the latest breakthrough by UCLA bioengineers in speech technology for people with disabilities.
» You may take issue with Ed Zitron’s skeptical perspective on AI, but he raises valid concerns and offers a necessary counterpoint to Silicon Valley’s potentially overly enthusiastic outlook on the future of large language models.
» By now, it’s abundantly clear who has emerged victorious in the startup accelerator arena—Y Combinator, having just raised a staggering $2 billion for its next two years of operations.
» A peculiar quirk of our modern era: Researchers forced to rename human genes to prevent Microsoft Excel from mistakenly interpreting them as dates.
» First we had Devin, now we get Devika—the open source cousin of the code-generating AI agent.
» Generative AI’s hype train derails as reality bites, but the journey is far from over.
Some Fun Stuff
🤔 Curious about the true impact of your LinkedIn profile picture? 🤳 Ponder no further! 🤖 Artificial intelligence shall reveal all. 😎
Some random fun links:
Enchanting particle simulation of a fluffy ball in your browser
Indulge in a few precious minutes popping some virtual bubble wrap
A nation for anti-agers—where pizza, donuts and alcohol will be illegal
Why don’t humans have tails? The Alu element might be the reason.
What actually happens to Google Maps when tectonic plates move?
Cat-GPT Proves That There’s a Glimmer of Hope for Every Startup Founder and Entrepreneur