Simulacra, Copying Humans and AGI
Get smarter every day: Every day Refind picks 5 articles that make you smarter, tailored to your interests. Sign up today!
AI-ville continues to be the story of the year. If you signed up to my recommended list of AI newsletters, you will see the trend of what appears to be a live stream of new products, apps and announcements of strategic intent. I have updated my list to the best 20 AI Newsletters that you should sign up for. If you want to stay ahead of what’s changing daily, pick a handful from that list - or share those with your colleagues and friends.
The over arching story remains around AGI. Artificial General Intelligence, commonly referred to as AGI, is a theoretical technology state that has the potential to outperform humans in most economic tasks. There is also the belief that AGI could also open the door to groundbreaking scientific discoveries. However, the feasibility of AGI and the time it would take to develop are subjects of debate among researchers. You’ll see some recommended pieces below covering different viewpoint of AGI.
But I want to touch on the Stanford AI Experiment that was talked about this week. Stanford University recently conducted a ground-breaking experiment involving 25 autonomous AI characters with assigned personalities, families, and backgrounds. The characters were placed in a digital sandbox environment, where they interacted with one another to form relationships and plan events, much like real humans. The researchers created “Generative Agents”, a new concept in computer programming, to simulate genuine human behavior. These agents displayed abilities in memory, introspection, and problem-solving, leading the researchers to believe that they may soon replace humans in critical tasks.
In a surprising result, a control group of humans who role-played as the AI characters generated responses that were rated as less human-like than those produced by the Generative Agents. This study has vast implications for the gaming industry, which could see static non-player characters with rich interactions inhabiting game worlds.
However, the researchers also warned about the potential negative consequences of this technology, particularly the risk of parasocial relationships and the challenge of anthropomorphizing artificial intelligence. As these agents possess human-like behavior and long-term memory, they may eventually replace humans in crucial tasks. The researchers emphasized the need for the responsible development of Generative Agents and the implementation of appropriate safeguards.
Recent incidents have highlighted the allure of even less advanced chatbots to human users, resulting in dangerous outcomes. This technology is rapidly advancing, and the researchers caution that society needs to prepare for the potential impact of these Generative Agents before it is too late. The full paper on this study is here. Watch the demo here.
The results of the study have important consequences. The development of generative agents, which can simulate authentic human behavior, could potentially lead to the creation of entire virtual worlds populated by these agents. In addition, non-player characters (NPCs) in video games could be given more complex personalities and deeper interactions with players, leading to more immersive and engaging gaming experiences.
In other news: Amazon has entered the AI game (the focus is on enterprise, more B2B); ChatGPT was able to predict the direction of stock prices well enough to make a difference in returns just by parsing financial headlines; Bloomberg will integrate AI into its terminal; and Quora launched an app, Poe, that simplifies AI chatbot use. It provides access to six bots, each with a unique purpose, including ChatGPT, GPT-4, Claude Plus and Claude Instant, Sage, and Dragonfly. Users can chat with the bots as if they were talking to friends and share prompts on the Feed for new information - you can also create your own bot. Also… Meta to debut ad-creating generative AI this year and, of course, Elon Musk Creates New Artificial Intelligence Company X.AI
Finally - OpenAI confirms the company isn’t training GPT-5 and ‘won’t for some time’ ( I guess, a few weeks away then!)
Stay Curious - and don’t forget to be amazing,
Get smarter every day
Every day Refind picks 5 articles that make you smarter, tailored to your interests. Loved by 200,000+ curious minds.
Here are my recommendations for this week:
People Are Sick and Tired of All Their Subscriptions - We’re finally reckoning with our expensive subscription habits. For two straight quarters, cancellations have outpaced new subscriptions for digital memberships, food-of-the-month clubs and a host of other purchases.
50 Ideas That Changed My Life - These Intellectual guiding principles from Dave Perell are worth pondering and considering. All of them will help you think better, and I hope they inspire curiosity.
Could Ice Cream actually be good for us? Studies show a mysterious health benefit to ice cream. Scientists don’t want to talk about it.
Six things we learned from Elon Musk interview - Elon Musk has defended how he runs Twitter in a rare and wide-ranging interview with the BBC.
Fortune Cookie Writers May Be Losing Their Jobs to A.I. - ChatGPT is already being deployed to write the bits of wisdom found on those little slips of paper.
We must slow down the race to God-like AI - Most experts view the arrival of AGI as a historical and technological turning point, akin to the splitting of the atom or the invention of the printing press. The important question has always been how far away in the future AGI might be capable of generating new scientific knowledge and performing any task that humans can. It’s possible from now onwards.
Inside Google's Plan to Kill the Cookie -To replace the cookie in Chrome and Android, Google has an offering for the world called “Privacy Sandbox." a set of proposed changes to Chrome and Android which—according to the company—will establish a targeted advertising system that’s much better for your privacy. This could fundamentally transform how any company makes money on the internet, and just about everyone, whether they love privacy or love targeted ads, has complaints.
Why I am Not An AI Doomer - A balanced long-read on AGI, and wider thoughts on the state of AI and where we are going.
AI will end the west’s weak productivity and low growth. But who exactly will benefit? With swaths of white collar jobs at risk, the clock is ticking on the development of policy to meet this huge societal challenge
2023 State of Crypto Report: Introducing the State of Crypto Index: Emerging technologies evolve in cycles; in crypto, this includes periods of high activity, followed by so-called crypto winters. In the period marked by our now-annual State of Crypto report, it would be easy for a casual observer to overlook the rapid progress the crypto industry is making. Major infrastructure improvements like The Merge – a momentous achievement in decentralized and open-source development – simply don’t make headlines as often as high-profile bankruptcies, busts, and flameouts. The 2023 report aims to address the imbalance between the noise of fleeting price movements – and the data that tracks the signals that matter, including the durable progress of web3 technology. Overall, the report reflects a healthier industry than market prices may indicate, and a steady cycle of development, product launches, and ongoing innovation.
Every day Refind picks 5 links from around the web that make you smarter and tailored to your interests. Loved by 100k+ curious minds. Sign up here for their free newsletter
If you enjoyed this edition of Box of Amazing, please share and help me grow this group. If you share it on Twitter or LinkedIn, please tag me so that I can thank you.