AI’s 2nd contact with humanity: are we losing again?

If the designers of an airplane told you that they estimated there was a 10% chance of the plane crashing, would you fly on it? This is where we are with current AI.

This month we’ll be discussing AI and its disruptive powers. Not the ‘singularity’ that Artificial General Intelligence is likely to bring, but what exists today.

The suggested input to our discussion is the messaging being put out by the Centre for Humane Technology, and specifically by its founder, Tristan Harris. Harris gets invited to the global meetings where CEOs of AI companies and politicians discuss how to globally manage AI. I feel his perspective is particularly helpful and clear.

Please watch this 42-minute video:
https://www.youtube.com/watch?v=e5dQ5zEuE9Q

I’ve provided a summary of the key points below, but please don’t use this as an alternative to the video, which includes many additional points and nuances and perspectives.

First exposure to AI: Curation AI

• Humanity’s first exposure to AI was via Social Media: AI curates the content that we see. Humanity lost as a result of that exposure.
• It promised (and delivered) real benefits, such as giving everybody a voice, the ability to connect with friends, help joining like-minded communities, help for small and medium-sized businesses to reach their customers and more.
• But the INCENTIVE behind it was The Race for Engagement (The race to engage the bottom of the brain stem) and this led to many harms that are continuing today and are not being addressed, such as…
• Addiction, doomscrolling, sexualization of kids, information overload, fake news, polarization, breakdown of democracy … and more
• These cannot be addressed individually - it requires the incentive to be changed, and there’s no obvious way to do that.

Second exposure to AI: Generative AI

• This second exposure comes from Large Language Models. The real power behind these is their (recent) ability to translate between multiple “languages” (forms of expression/data representation), such as text and images, brain signals and images, text and psychological models etc.
• Generative AI promised (and is delivering) greater efficiency, faster coding, finding cures for cancer, helping us solve climate change, increasing GDP (assuming that’s a good thing - but that’s another discussion) - and much, much more
• But the INCENTIVE behind it is The Race to Deploy Impressive Capabilities as Quickly As Possible and that is creating many new and deeper societal dysfunctions, including..
• … automated bioweapons, automated cyberweapons, exponential fraud & crime, exponential fake and child porn, institutional overwhelm, destabilization of nation states and much more.
• As with Curation AI, there is no possibility of addressing each of these Generative AI issues individually: it’s the INCENTIVE that needs to be addressed.

Tristan Harris indicates that all the players are aware of the problems and there are meetings and discussion going on - but as far as I can see, there seems little evidence that concrete actions will happen that will address the incentive.

Questions we could discuss include:

• Is there any reason to distrust Harris's analysis?
• What are the systemic barriers to changing the INCENTIVE?
• … and what would it take to dismantle them?
• If the answer is ‘there is almost nothing we can do to change the incentive’ - then what can we do to protect ourselves, and those close to us, from the dysfunctions that target individuals? (And here, we could include the dysfunctions caused by Curation AI as well.)
• Are there ways of using AI itself to support such protections?

Note: we are not in our usual venue. Seems there is too much festivity at the Red Lion.

Keywords
humanity, artificial intelligence, environmental awareness, politics, philosophy, spiritual growth, economics, ecological economics, intentional communities, geopolitics, open sources, the commons, activism, idealism, capitalism, sustainable agriculture, religion, consciousness, inequality, enlightenment, education, autocracy, corporatism, metamodernism