please In the loopTime's two newsletters on the world of AI. If you're reading this in your browser, you can subscribe to deliver the following directly to your inbox:
Subscribe to In the loop
What you know: AI Social Contract
Earlier this month, 18 people from Openai, Google Deepmind, the UK Institute of AI Security, OECD and other groups gathered for an invitation-only summit. On the agenda: We arrive at understanding how the ways in which AI may have advanced can affect the “social contract” between workers, governments, and businesses.
Top AI CEOs like Deepmind's Demis Hassabis and Openai's Sam Altman have recently urged academics and governments to tackle this issue more deeply, urging the world to better prepare themselves to meet what they expect. So, every day, every day in breakout rooms and nightly communal saunas – these experts hid pictures of what economic shocks are emerging and what to do about them.
Bad news – According to the summit organizers, the results of the so-called “AGI Social Contract Summit” were a list of four draft statements. These statements have not been reported previously. They paint a harsh picture of where the world is heading without any serious intervention by governments or society. “AI can exacerbate the rise in wealth and income inequality within the country and worsen the economic situation for many workers, middle class people and families,” reads initially. “AI increases inequality between countries with and without access to AI infrastructure. In terms of benefits and ability to respond to shocks,” the second says. “Without intervention, AI response inequality could lead to political control of wealthy individuals and businesses, erode democratic institutions, and increase levels of political dissatisfaction,” the third says. And fourth: “The intrusion of AI systems and erosion of the value of labor leads to an increase in the impotence of most humans, leading to the deterioration of individual happiness and purpose.”
Human dismantling – Summit attendees agreed that existing social contracts where people receive security and social interests in exchange for labor are problematic for AI, according to Deric Cheng, the event organizer, who is research director for Windfall Trust, founded this year to tackle these issues. “We are essentially concerned that labor will make it somewhat easier for a government to be neutralised compared to businesses and to some extent, in some degree,” Chen says. “The obvious consequence of the decline in the labor force is a decline in real wages.” In this view, wealthy democratic people enjoy a high standard of living, not because of their paper-enforced rights. Even if the labor force is removed from that equation and the overall GDP or productivity statistics increase, living standards are vulnerable to downwards.
Progress – Without government intervention, participants agreed, but the default path of advanced AI is likely to result in poor economic outcomes for the average person. But fortunately, Chen says they have identified some possible actions that the government can take to push things up for the better. For example, we develop new institutions in the IMF vein to ensure that wealth derived from AI is distributed globally, rather than one or two powerful domestics in which AI companies are located. The state can also run pilots today, Cheng said, and they can also gather evidence on what types of safety nets are effective due to policies such as basic income and shortened workweeks.
Google Deepmind declined to comment on the statement arising from the summit. Openai did not respond to requests for comment. After the publication of this article, Cheng said in the message that all participants existed in personal abilities rather than formal abilities, and that the statement did not reflect the views of the organization. He also added that some attendees disagree with some of the draft statements that they previously described as “consensus” of the group.
With one minute, take part in a quick research so you can better understand who you are and which AI topics are most interested in.
Please participate in the survey
People you need to know: US District Judge Amit Meta
Last year, Federal District Judge Amit Mehta ruled that Google had maintained its monopoly on online search and advertising illegally. This week he is expected to announce a court decision on what to do about it. This could range from sharing Google's shared data with rivals to forcing the disbandment of the search giant itself.
Payment to rivals – The U.S. Department of Justice lawsuit against Google revolved around the billions of dollars Google made to Apple to protect it as the iPhone's default search engine. Observers expect the court to at least limit these types of payments.
Chrome Spinning – Another possibility is that Mehta can order Google to sell Chrome, the world's most popular browser, with a 67% market share. With Chrome, Google can collect complex data about user browsing patterns that enhance control over search and advertising spaces. Given the opportunity for one of Google's competitors to refer to the LLM of their choice, there's no doubt that they'll jump to the opportunity to buy the world's top browsers.
Sharing user data – The data that Google collects with users is part of the secret source of search engines. Mehta can determine that Google needs to share this data with its competitors. It could probably be anonymized form to prevent accusations of privacy violations.
AI Action
Public trusts AI chatbots more than corporate and community leaders, according to a vote of users from 68 countries conducted by the Collective Intelligence Project.
More than half (56.6%) of votes believe they trust AI chatbots. This is higher than AI companies (34.6%) or AI companies that create faith and community leaders (44.2%).
According to the survey, more than one in 10 people (14.9%) use AI for emotional support. And 30% of people “at some point they thought the AI chatbot was self-aware.”
And 56% of those who voted said that a surge in AI across society is likely to worsen access to good jobs.
As always, if you have an interesting story about AI behavior, we'd love to hear it. Please contact us by email: intheloop@time.com
What we're reading
Competition for artificial general information poses new risks to an unstable world with Billy Perrigo
Shameless plug for my own story here. Earlier this year, I traveled to Paris to take part in a fascinating exercise. In the simulated war game, four teams showed off the impact of advanced AI on geopolitics. It was like watching the dungeon and dragon games, except that the players were former government officials and AI researchers, and the game board was planet Earth. I use war games as a jump-off point in the story to explore how artificial general information has become an increasingly prominent dimension of the massive competition between the US and China. I hope you read it and let me know what you think!
Corrected, August 30th
The original version of this article mischaracterized the level of agreement among summit participants. After the release, Chen said that some attendees were against some statements, not a true consensus. Additionally, this article has been updated to note that all participants exist in personal abilities rather than formal abilities, and that the statement does not reflect the views of the organization.
