Fluidity

This concludes "Gradient Dissent", the companion document to "Better Without AI". Thank you so much for listening! You can support the podcast and get episodes a week early, by supporting the Patreon: https://www.patreon.com/m/fluidityaudiobooks   If you like the show, consider buying me a coffee: https://www.buymeacoffee.com/mattarnold   Original music by Kevin MacLeod.   This podcast is under a...

Current text generators, such as ChatGPT, are highly unreliable, difficult to use effectively, unable to do many things we might want them to, and extremely expensive to develop and run. These defects are inherent in their underlying technology. Quite different methods could plausibly remedy all these defects. Would that be good, or bad? https://betterwithout.ai/better-text-generators John McCarth...

Analysis of image classifiers demonstrates that it is possible to understand backprop networks at the task-relevant run-time algorithmic level. In these systems, at least, networks gain their power from deploying massive parallelism to check for the presence of a vast number of simple, shallow patterns. https://betterwithout.ai/images-surface-features This episode has a lot of links: David Chapman...

Current AI practice is not engineering, even when it aims for practical applications, because it is not based on scientific understanding. Enforcing engineering norms on the field could lead to considerably safer systems.   https://betterwithout.ai/AI-as-engineering   This episode has a lot of links! Here they are.   Michael Nielsen's "The role of 'explanation' in AI". https://michaelnotebook.com/...

Few AI experiments constitute meaningful tests of hypotheses. As a branch of machine learning research, AI science has concentrated on black box investigation of training time phenomena. The best of this work is has been scientifically excellent. However, the hypotheses tested are mainly irrelevant to user and societal concerns. https://betterwithout.ai/AI-as-science This chapter references Chapma...

Do AI As Science And Engineering Instead - We've seen that current AI practice leads to technologies that are expensive, difficult to apply in real-world situations, and inherently unsafe. Neglected scientific and engineering investigations can bring better understanding of the risks of current AI technology, and can lead to safer technologies.   https://betterwithout.ai/science-engineering-vs-AI...

Current AI results from experimental variation of mechanisms, unguided by theoretical principles. That has produced systems that can do amazing things. On the other hand, they are extremely error-prone and therefore unsafe. Backpropaganda, a collection of misleading ways of talking about "neural networks," justifies continuing in this misguided direction.   https://betterwithout.ai/backpropaganda...

The conclusion of this chapter.   So-called "neural networks" are extremely expensive, poorly understood, unfixably unreliable, deceptive, data hungry, and inherently limited in capabilities. In short: they are bad.  https://betterwithout.ai/artificial-neurons-considered-harmful Sayash Kapoor and Arvind Narayanan's "The bait and switch behind AI risk prediction tools": https://aisnakeoil.substack....

This begins "Gradient Dissent", the companion material to "Better Without AI". The neural network and GPT technologies that power current artificial intelligence are exceptionally error prone, deceptive, poorly understood, and dangerous. They are widely used without adequate safeguards in situations where they cause increasing harms. They are not inevitable, and we should replace them with better...

The five short chapters in this episode are the conclusion of the main body of Better Without AI. Next, we'll begin the book's appendix, Gradient Dissent. Cozy Futurism - If we knew we'd never get flying cars, most people wouldn't care. What do we care about? https://betterwithout.ai/cozy-futurism Meaningful Futurism - Likeable futures are meaningful, not just materially comfortable. Bringing one...

A Future We Would Like - The most important questions are not about technology but about us. What sorts of future would we like? What role could AI play in getting us there, and also in that world? What is your own role in helping that happen? https://betterwithout.ai/a-future-we-would-like How AI Destroyed The Future -We are doing a terrible job of thinking about the most important question becau...

Stop obstructing scientific progress! We already know how to dramatically accelerate science: by getting out of the way.   https://betterwithout.ai/stop-obstructing-science   How to science better. What do exceptional scientists do differently from mediocre ones? Can we train currently-mediocre ones to do better?   https://betterwithout.ai/better-science-without-AI   Scenius: upgrading science FTW...

Progress requires experimentation. Suggested ways AI could speed progress by automating experiments appear mistaken. https://betterwithout.ai/limits-to-induction You can support the podcast and get episodes a week early, by supporting the Patreon: https://www.patreon.com/m/fluidityaudiobooks If you like the show, consider buying me a coffee: https://www.buymeacoffee.com/mattarnold Original music b...

Forgive the sound quality on this episode; I recorded it live in front of an audience on a platform floating in a lake during the 2024 solar eclipse. This is a standalone essay by David Chapman on metarationaity.com. How scientific research is like cunnilingus: a phenomenology of epistemology. https://metarationality.com/going-down-on-the-phenomenon You can support the podcast and get episodes a w...

What Is The Role Of Intelligence In Science? Actually, what are "science" and "intelligence"? Precise, explicit definitions aren't necessary, but discussions of Transformative AI seem to depend implicitly on particular models of both. It matters if those models are wrong. https://betterwithout.ai/intelligence-in-science Katja Grace, "Counterarguments to the basic AI x-risk case". https://aiimpacts...

Radical Progress Without Scary AI: Technological progress, in medicine for example, provides an altruistic motivation for developing more powerful AIs. I suggest that AI may be unnecessary, or even irrelevant, for that. We may be able to get the benefits without the risks. https://betterwithout.ai/radical-progress-without-AI What kind of AI might accelerate technological progress?: "Narrow" AI sys...

Recognize that AI is probably net harmful: Actually-existing and near-future AIs are net harmful—never mind their longer-term risks. We should shut them down, not pussyfoot around hoping they can somehow be made safe. https://betterwithout.ai/AI-is-harmful Create a negative public image for AI: Most funding for AI research comes from the advertising industry. Their primary motivation may be to cre...

"Apocalypse now" identified the corrosive influence of new viral ideologies, created unintentionally by recommender systems, as a major AI risk. These may cause social collapse if not tackled head-on. You can resist. https://betterwithout.ai/spurn-artificial-ideology Announcement tweet for the Opening Awareness, Opening Rationality discussion group starting on February 1:  https://twitter.com/open...

Current AI practices produce technologies that are expensive, difficult to apply in real-world situations, and inherently unsafe. Neglected scientific and engineering investigations can bring better understanding of specific risks of current AI technology, and can lead to safer technologies.   https://betterwithout.ai/fight-unsafe-AI You can support the podcast and get episodes a week early, by su...

The technologies underlying current AI systems are inherently, unfixably unreliable. They should be deprecated, avoided, regulated, and replaced. https://betterwithout.ai/mistrust-machine-learning You can support the podcast and get episodes a week early, by supporting the Patreon: https://www.patreon.com/m/fluidityaudiobooks If you like the show, consider buying me a coffee: https://www.buymeacof...

Gaining unauthorized access to computer systems is a key source of power in many AI doom scenarios. That is easy now, because there are scant incentives for serious cybersecurity; so nearly all systems are radically insecure. Technical and political initiatives must mitigate this problem. https://betterwithout.ai/cybersecurity-vs-AI   You can support the podcast and get episodes a week early, by s...

Practical Actions You Can Take Against AI Risks: We can and should protect against current and likely future harmful AI effects. This chapter recommends practical, near-term risk reduction measures. I suggest actions for the general public, computer professionals, AI ethics and safety organizations, funders, and governments. https://betterwithout.ai/pragmatic-AI-safety End Digital Surveillance: Da...

This concludes the "Apocalypse Now" section of Better Without AI.   AI systems may cause near-term disasters through their proven ability to shatter societies and cultures. These might potentially cause human extinction, but are more likely to scale up to the level of the twentieth century dictatorships, genocides, and world wars. It would be wise to anticipate possible harms in as much detail as...

Who is in control of AI? - It may already be too late to shut down the existing AI systems that could destroy civilization. https://betterwithout.ai/AI-is-out-of-control What an AI apocalypse may look like - Scenarios in which artificial intelligence systems degrade critical institutions to the point of collapse seem to me not just likely, but well under way. https://betterwithout.ai/AI-safety-fai...

"In this audiobook... A LARGE BOLD FONT IN ALL CAPITAL LETTERS SOUNDS LIKE THIS."   Apocalypse now - Current AI systems are already harmful. They pose apocalyptic risks even without further technology development. This chapter explains why; explores a possible path for near-term human extinction via AI; and sketches several disaster scenarios.   https://betterwithout.ai/apocalypse-now   At war wit...