#109

Рет қаралды 13,476

Күн бұрын

Pod version: podcasters.spotify.com/pod/sh...
Support us! / mlst
MLST Discord: / discord
Dan McQuillan, a visionary in digital culture and social innovation, emphasizes the importance of understanding technology's complex relationship with society. As an academic at Goldsmiths, University of London, he fosters interdisciplinary collaboration and champions data-driven equity and ethical technology. Dan's career includes roles at Amnesty International and Social Innovation Camp, showcasing technology's potential to empower and bring about positive change. In this conversation, we discuss the challenges and opportunities at the intersection of technology and society, exploring the profound impact of our digital world.
Interviewer: Dr. Tim Scarfe
[00:00:00] Intro
[00:02:20] Dan's background and journey to academia
[00:04:10] Writing the book "Resisting AI"
[00:08:30] Necropolitics and its relation to AI
[00:10:06] AI as a new form of colonization
[00:12:57] LLMs as a new form of neo-techno-imperialism
[00:15:47] Technology for good and AGI's skewed worldview
[00:17:49] Transhumanism, eugenics, and intelligence
[00:20:45] Valuing differences (disability) and challenging societal norms
[00:26:08] Re-ontologizing and the philosophy of information
[00:28:19] New materialism and the impact of technology on society
[00:30:32] Intelligence, meaning, and materiality
[00:31:43] The constraints of physical laws and the importance of science
[00:32:44] Exploring possibilities to reduce suffering and increase well-being
[00:33:29] The division between meaning and material in our experiences
[00:35:36] Machine learning, data science, and neoplatonic approach to understanding reality
[00:37:56] Different understandings of cognition, thought, and consciousness
[00:39:15] Enactivism and its variants in cognitive science
[00:40:58] Jordan Peterson
[00:44:47] Relationism, relativism, and finding the correct relational framework
[00:47:42] Recognizing privilege and its impact on social interactions
[00:49:10] Intersectionality / Feminist thinking and the concept of care in social structures
[00:51:46] Intersectionality and its role in understanding social inequalities
[00:54:26] The entanglement of history, technology, and politics
[00:57:39] ChatGPT article - we come to bury ChatGPT
[00:59:41] Statistical pattern learning and convincing patterns in AI
[01:01:27] Anthropomorphization and understanding in AI
[01:03:26] AI in education and critical thinking
[01:06:09] European Union policies and trustable AI
[01:07:52] AI reliability and the halo effect
[01:09:26] AI as a tool enmeshed in society
[01:13:49] Luddites
[01:14:49] AI and systemic issues
[01:15:16] AI is a scam
[01:15:31] AI and Social Relations
[01:16:49] Invisible Labor in AI and Machine Learning
[01:18:38] The Need for Human Care in AI Implementation
[01:21:09] Exploititative AI / alignment
[01:23:50] Science fiction AI / moral frameworks
[01:27:22] Discussing Stochastic Parrots and Nihilism
[01:30:36] Human Intelligence vs. Language Models
[01:32:22] Image Recognition and Emulation vs. Experience
[01:34:32] Thought Experiments and Philosophy in AI Ethics (mimicry)
[01:38:42] The Trolley Problem and Self-Driving Cars
[01:41:23] Abstraction, reduction, and grounding in reality
[01:43:13] Process philosophy and the possibility of change
[01:49:55] Mental health, AI, and epistemic injustice
[01:50:30] Hermeneutic injustice and gendered techniques
[01:52:03] Machine learning and welfare fraud detection
[01:53:57] AI and politics
[01:55:28] Random forests and Goodhart's law
[01:59:24] Epistemic injustice and testimonial injustice
[02:03:02] Suspicion and socially stratified experiences
[02:07:39] Isnt language oppressive
[02:10:51] YT comment rant!
[02:11:46] Fascism and AI discussion
[02:13:24] Violence in various systems
[02:16:52] Recognising systemic violence
[02:19:14] Technical systems and feedback loops
[02:21:41] Language and its impact on understanding violence
[02:22:35] Fascism in Today's Society
[02:27:35] Defining Fascism and its Emergence
[02:32:13] The Law and its Limitations
[02:33:33] Pace and Scale of Technological Change
[02:37:38] Alternative approaches to AI and society
[02:40:44] Unintelligible Beginnings and Pragmatic Approaches
[02:42:38] Algorithmic Systems and AI in Decision-Making
[02:44:09] Self-Organization at Successive Scales / cybernetics
[02:46:26] The Importance of Pluralism and Local Expertise
[02:48:24] Eliminating subversive elements
References are in the pinned comment

Пікірлер: 177

@MachineLearningStreetTalk Жыл бұрын

References: AI, Ethics, and Philosophy Dan McQuillan / LinkedIn www.linkedin.com/in/danmcquillan/ twitter.com/danmcquillan Data Science as Machinic Neoplatonism (Dan McQuill) link.springer.com/article/10.1007/s13347-017-0273-3 We come to bury ChatGPT, not to praise it. (Dan McQuill) www.danmcquillan.org/chatgpt.html Goldsmiths University of London / Website www.gold.ac.uk/ Resisting AI by Dan McQuillan / Bristol University Press bristoluniversitypress.co.uk/resisting-ai Akela Mbembe / Wikipedia en.wikipedia.org/wiki/Achille_Mbembe Sam Altman / Twitter twitter.com/sama OpenAI / Website openai.com/ Noam Chomsky / Wikipedia en.wikipedia.org/wiki/Noam_Chomsky AI Researchers and Thinkers Effective Altruism / Movement www.effectivealtruism.org/ Michael Levin / Biologist and researcher ase.tufts.edu/biology/labs/levin/ Luciano Floridi / Philosopher and Professor www.philosophy.ox.ac.uk/people/luciano-floridi Gary Marcus / Cognitive scientist, author, and entrepreneur garymarcus.substack.com/ Geoffrey Hinton / AI researcher and pioneer in deep learning www.cs.toronto.edu/~hinton/ Authors and Philosophers Philosophy Tube / Abigail Thorne kzfaq.info/love/2PA-AKmVpU6NKCGtZq_rKQ Donna Haraway / The Cyborg Manifesto en.wikipedia.org/wiki/The_Cyborg_Manifesto Jordan Peterson / Maps of Meaning www.amazon.com/Maps-Meaning-Architecture-Jordan-Peterson/dp/0415922224 Kate Crawford / Atlas of AI www.amazon.co.uk/Atlas-AI-Kate-Crawford/dp/0300209576 AI and Society Topics Emily Bender / Twitter twitter.com/emilymbender Searle's Chinese Room Argument / Stanford Encyclopedia of Philosophy plato.stanford.edu/entries/chinese-room/ David Chalmers / Philosophical Zombie plato.stanford.edu/entries/zombies/ Descartes / Mind-Body Dualism plato.stanford.edu/entries/dualism/#CarSub Roe vs. Wade Decision / Oyez www.oyez.org/cases/1971/70-18 Luciano Floridi / Oxford Internet Institute www.oii.ox.ac.uk/people/luciano-floridi/ Disability Movement / "Nothing About Us Without Us" en.wikipedia.org/wiki/Nothing_About_Us_Without_Us Miranda Fricker / Epistemic Injustice: Power and the Ethics of Knowing academic.oup.com/book/32817 Lighthouse Reports / Investigative journalism organization www.example.com Concepts and Theories Random Forest Fraud Detection / Random Forest Algorithm en.wikipedia.org/wiki/Random_forest Goodhart's Law en.wikipedia.org/wiki/Goodhart%27s_law Epistemic Injustice en.wikipedia.org/wiki/Epistemic_injustice Testimonial Injustice en.wikipedia.org/wiki/Testimonial_injustice Hannah Arendt en.wikipedia.org/wiki/Hannah_Arendt Chomsky on language models / Noam Chomsky on AI kzfaq.info/get/bejne/l96lesmZmbu2aYU.html Books and Conferences Universal Credit / UK government benefit system www.gov.uk/universal-credit Fascist Governments and Policies Fascist Prime Minister in Italy / Giuseppe Conte en.wikipedia.org/wiki/Giuseppe_Conte Fascistic Governments in Hungary / Viktor Orbán en.wikipedia.org/wiki/Viktor_Orb%C3%A1n Agamben on Zones of Exception / Giorgio Agamben en.wikipedia.org/wiki/Giorgio_Agamben#State_of_Exception Cybernetics and Alternative Plans Andrew Pickering's Book on Lost Cybernetics / The Cybernetic Brain press.uchicago.edu/ucp/books/book/chicago/C/bo8169881.html Stafford Beer's Cybernetics Model en.wikipedia.org/wiki/Stafford_Beer Karl Friston's Free Energy Principle en.wikipedia.org/wiki/Free_energy_principle Lucas Plan en.wikipedia.org/wiki/Lucas_Plan Noam Chomsky's Views on Corporations chomsky.info/1995____/

@MachineLearningStreetTalk Жыл бұрын

Note that these opinions are not Tim's and that MLST seeks to have respectful conversations with folks from across the spectrum. We are extremely gracious that Dan accepted our invitation to come on MLST, and is the first expert in this area who actually has (not for the want of trying!).

@cacogenicist Жыл бұрын

You can be respectful while still probing (what I found to be quite a bit of nonsense) a bit more energetically than was done here.

@lkyuvsad Жыл бұрын

19:02 is critical I spent most of my early adult life in a similar philosophical milieu to the one you'd find in silicon valley. I was knocked out of my bubble by a series of ongoing tragedies, which gave me the opportunity to spend time outside my monoculture of self-satisfied, academically bright people with adequate disposable income. I feel a deep sense of embarrassment that I so over-valued the same kind of intellectual capacities I see rationalists and trans-humanists also focused on. There is so much more that is precious about humans than our productive, intellectual or even creative capacity (not that the latter are without value). Every one of the insights I treasure most from the last 20 years of my life I had to be taught by kinder people than myself with a fraction of the academic capability. Smart people can be so stupid. Rationalists talk a lot about our need for more geniuses. We are perhaps more desperately in need of compassionate, caring, humble people.

@annjuurinen6553 Жыл бұрын

Lovely thoughts. So true. I hope you are recovering from your dilemmas. You have learned so much.

@caneprints Жыл бұрын

As someone who has been totally blind from birth, AI has helped me in so many ways, including giving me descriptions of things I cannot see. The unfortunate reality is that human beings always seem to want to find new ways to destroy each other, and of course AI will be used for that purpose I am sure. I don’t think there’s a solution to that problem unless you can somehow change the human brain. With all the good things AI can do, it is such a shame that we have to spend so much time and effort focusing on the negatives. Aside from destroying ourselves as a species, which we have been perfecting for as long as we have existed, my biggest fear is that humans will delegate so much of their thinking to machines that we will have nothing productive left to do. I mean, if you were a parent, what kind of career advice would you give to a young person when AI seems to be taking over everything? there is a radio station which is now totally autonomous, with AI DJs and even AI music compositions. AI can generate art so much better than people, so what is the point in humans trying to generate art at all? Maybe this question will finally be answered when rich people can no longer sell their goods and services to the rest of us because we are all poor and unemployed.

@Pslam23JesusIsLord Жыл бұрын

Lord Jesus Christ is returning! Be prepared for His Return, everyone! He loves you very much, accept Him in your Heart and believe on Him, and He is coming for His Bride. There is a shift and we all can feel it. It happened in the spiritual and it is going to manifest in the physical!

@Ulyssestnt Жыл бұрын

That's a profound and well thought out answer,thank you. I'm glad to hear AI has augmented your life as it have mine. There are always large upsets in society when new revolutionary technologies are introduced ,I wen through this in the finance industry a decade plus ago when narrow AI took over the open outcry floors. The catholic church wanted desperately to stop the rollout of the printing press hundreds of years ago. The technology will augment humans and it's not magic..it's made by humans and it's a tool like any other tool in our toolkit.

@StoutProper Жыл бұрын

AI is already being used to that extent massively in Ukraine.

@MachineLearningStreetTalk Жыл бұрын

Sorry if the TOC is a bit verbose, here is a more zoomed out one: [00:00:00] Early discussion about AI and society [00:15:00] AI and human values [00:30:00] Intelligence, meaning, and materiality [00:45:00] AI and social relations [01:00:00] AI ethics and policy [01:15:00] AI and systemic issues [01:30:00] AI and philosophy [01:45:00] AI and the law [02:00:00] AI and politics [02:15:00] AI and Fascism [02:30:00] Fascism in Today's Society [02:45:00] Alternative approaches to AI and society

@robertomonroe6338 Жыл бұрын

This is vital work which investigates the role of algorithmic systems in political economy. Researchers, practioners and enthusiasts talk about 'AI' as if it's like gravity, unmoored from the materiality of computation (resource extraction, supply chains) and with little regard for social impacts. Dr. McQuillan re-centers this. which makes many uncomfortable.

@StoutProper Жыл бұрын

Totally agree. And while you don’t need men in dark rooms planning to have a conspiracy, there are many locuses directing the response against subversive elements, the CIA, FBI and MI5 for a start. Why else would they have members on the boards of Twitter and Facebook and want to ban Tiktok?

@peternguyen2022 Жыл бұрын

Thanks Tim for a great interview. Listening to Dr McQuillan, I remembered a line in the movie Flowers for Algernon where the new Charlie Gordon says, "In (the old) Charlie Gordon's world, we are the morons." Is humanity in danger of following the same path as Charlie Gordon where we become smarter and smarter ("smart" being defined in a very rigid and stringent manner), yet we forget the intelligence of our heart? In the 2000 film adaptation of Flowers for Algernon, we see the original Charlie as a warm, caring and friendly human being. I'm not against AGI, but I would agree with Dr McQuillan that perhaps a more central (albeit slightly embarrassing) question is, "Why is it that we don't care about others?" My first reaction is, "Well, most people (including myself) are looking out for number one. That is, they must first ensure their own social/economic/financial survival first before they can care about strangers." If that is the case, then perhaps AI (like ChatGPT) can help people make MORE money in LESS time so they do have time to devote to the community as volunteers. But then again, human nature being what it is, perhaps AI-enhanced professionals/entrepreneurs will end up being more motivated than ever to make even more money -- that is, someone who didn't care for others is perhaps unlikely to care more when they earn more money. This is why this interview made such an impression on me. While nearly everyone else is talking about apocalyptic or utopian scenarios about AI, Dr McQuillan is asking a very simple yet urgent question: "Why don't we care about the suffering of others?"

@StoutProper Жыл бұрын

Perhaps it’s a question for AI: “In our society, the emphasis is often placed on individual success and profit, which can lead to a lack of concern for others who may not be able to contribute to the same level of success or profit. This focus on self-interest can create a culture of competition rather than collaboration, where individuals prioritize their own goals over the well-being of others. Additionally, in today’s society, resources and opportunities are often distributed unequally, which can lead to further disparities and a lack empathy for those who may not have the same advantages. This influences various other factors, including cultural upbringing, personal experiences, and individual differences, which together can lead to societal and structural factors such as discrimination, inequality, and lack of access to resources, that all contribute to a lack of care for others.”

@KeenAesthetic1 Жыл бұрын

I worked in Dan's team at Amnesty just before he left and wondered what he was doing, so great to see what he's come up with.

@oncedidactic Жыл бұрын

Bringing this perspective, and lucidly engaging it, is so valuable. It is the purpose of the mind to stop and think (and feel), to sense that which cannot be seen.

@markpovell Жыл бұрын

This for me is one of the most useful, insightful and informative contributions to the broader conversation about so called Artificial Intellingence that I have come across. Third parse and still 'processing' ha ha! Thank you both and thank you MLST for this Channel.

@roseproctor3177 7 ай бұрын

one of my favorite and most challenging episodes youve ever put out

@BrianPeiris Жыл бұрын

Thank you for inviting this perspective.

@akakico Жыл бұрын

Great conversation. Disagree with a lot of what Dan McQuillan was saying, but a great conversation.

@missadventuresmotorcycledi2773 Жыл бұрын

I would find it extraordinary not to agree entirely with Dan McQuillan.

@XOPOIIIO Жыл бұрын

I do not fear AI that has distorted vision of the world, I fear the AI that has accurate knowledge about the world and can manipulate it precisely.

@CodexPermutatio Жыл бұрын

If I understand you correctly, what you fear is an unaligned superintelligent AI making changes in the world in its own benefit. But a similar scenario could occur if an AI with a distorted view of reality is used by people as an assistant to make vital decisions. It would be an indirect way of manipulating the world (manipulating us, although not with "malicious" intent). The distorted AI might even "think" that it's doing good and helping us but, nevertheless, the unintended consequences might lead to a disaster. In my opinion, we should never delegate our most important decisions to an AI especially if we don't already have a way to ensure an alignment of such AI's behavior with fundamental human ethical principles.

@Morimea Жыл бұрын

Only largest corporations can play with AI, and they will use it to make money. And you will have absolutely nothing from it.

@sooraj1104 Жыл бұрын

You have to worry about both.

@TeodoraPetkova Жыл бұрын

Thank you so much for this wonderful dialogue! So much needed in the increasingly noisy information "ocean".

@missadventuresmotorcycledi2773 Жыл бұрын

A noisy ocean of men predominantly, now why is that?

@Rockyzach88 Жыл бұрын

I'm glad people are thinking about it, but I think some of the feelings expressed in this video are too reactionary. It seems older people always seem to be "over cautious" or reactionary towards technology. Maybe that's the only way they can pass their wisdom forward though. But we have examples in history of people saying these kinds of thing before. Some of those are the same issues, some are different. Some were never solved. I personally see it as a tool for further production and creativity. I don't see it as laziness whatsoever. Saying that its laziness to me tells me that people think that our current productivity, creativity, and various abstractions of society are the ceiling. I think that's when younger people come in and they are more capable of seeing greater abstractions from advancements in technology. Another thing is the "AI effect". "This AI is just this or it's just that". I wonder if this is exacerbated by the idea many have that humans are divine creations and not "just this or just that". Overall it was a thought provoking video though.

@wonmoreminute Жыл бұрын

I’m also fascinated by the amount of people who say “it’s just this or it’s just that”, or talk about it’s limitations as if it’s as good as it’ll ever get. I don’t know if it’s linear thinking, as if improvements only come gradually over years and decades (instead of weeks), or if it’s not having first-hand experience with it (particularly GPT-4) and just repeating what they’ve heard or read somewhere else. Or is it genuine insecurity and fear that maybe we as humans are not as impressive as we think we are. In many ways we’re just predicting what the next word is too, whether we’re speaking it, writing it, or thinking it. And through language we form thoughts. Of course, we think there's more going on and maybe when you throw in some chemical combinations for emotions and sensory inputs for pain and pleasure, we are more complex. But are we as complex as we've always assumed we are? Interestingly, Sam Altman and Lex Fridman on Lex’s recent podcast were not entirely certain ChatGPT wasn’t sentient. Lex hesitated when asked the question, and Sam said he didn’t know, but that it was interesting they were even having the conversation. I’ve been working with GPT-4 daily since it became available and no one can tell me it’s not intelligent. We "humans" can debate the definition of intelligence or even what it means to be sentient, but GPT-4 understands complex ideas and nuance. I can engage with it about deep and intellectual concepts, or simply ask it to perform complex tasks in a particular way, and it gets it. It doesn't matter if it's a "trick", it still does it. And again, maybe it's a trick when we do it. I also don’t need to be clever with prompts like I did with previous versions. GPT-4 understands what I want just by asking it. I haven’t really found it to be factually inaccurate unless I’m purposely trying to fool it. Factually incomplete maybe, so I'll ask it to elaborate on an idea. But quite often if it doesn’t know something, it’ll say it doesn’t know. Those who keep pointing at it’s hallucinations as if it’s a catastrophic failure of AI are clearly not using it on a regular basis and are talking about older versions. Which is understandable. In the context of AI, older means weeks ago. It occasionally gets some code wrong but when I explain what happens when I run it, it knows exactly what went wrong and it fixes it. Also, from what I’ve read, it’s capable of doing things it was never taught, and even OpenAI doesn't understand how or why. Not to mention, it can pass extremely difficult tests in multiple disciplines that require years of schooling, and it can do it with a 90 percent or more. So it is way more than “just this or just that” and I don’t think people comprehend how much the world has changed in the last few months. For example, it can easily do the work of 2 or 3 employees for me, and it can do it 10X faster. But let’s just say it’s good enough for just one employee. That means, if 10 million people use it as an assistant, OpenAI has essentially added 10 million “workers” into the labor force at a cost of pennies per worker. Imagine when 100 million, or 1 billion people are using it to 10X their productivity. And there will be use cases that we can’t even yet wrap our heads around. I should also say that none of my thoughts about this are with excitement. Fascination and genuine disbelief that this is where we are. But I'm worried about my own relevance too and in many ways, would love to go back to a time long before any of this. But I must also admit, when ChatGPT went down a few days ago, I was also somewhat lost without it. That's scary.

@flickwtchr Жыл бұрын

@@wonmoreminute "That means, if 10 million people use it as an assistant, OpenAI has essentially added 10 million “workers” into the labor force at a cost of pennies per worker." Displacing…

@CandidDate Жыл бұрын

A thought experiment. Ask a robot with a GPT-4 brain to pour a cup of water. There are two threads going on, the physical and the language describing the physical. So it says, "Pick up the pitcher of water with the correct pressure on the hands to hold its weight because gravity will want the pitcher to fall out of your hands." It does this. "Now tilt the pitcher over the glass and gravity will make a stream of water pour out of the pitcher." So it does this. "Make sure you time the pour so you don't pour too much or too little." A little calculus rate formula calculates how long to pour. "Straighten the pitcher, the water will stop pouring and set the pitcher down by releasing the pressure on the pitcher with my hands when the pitcher is upright and stable on the table." -- Do you see how language can describe just about any action we physically take in this world? In this way, and with a little math robots can do just about anything. The key is grounding of course, which is achieved by the visual input being correctly tied to the language.

@conan_der_barbar Жыл бұрын

using LLMs to teach robots using natural language is actually already being done in research

@hdjwkrbrnflfnfbrjrmd Жыл бұрын

@@conan_der_barbar i throw stones at homeless bums

@laurenpinschannels Жыл бұрын

I'm quite worried about being replaced by ai if some of the current research threads pan out. how can models work to be an augment for all humans, in a way entirely designed by the human using it?

@novelspace Жыл бұрын

Aggressive(in a good way that makes you think) and Insightful! I like the perspective of using it at a tool not intelligence. Ie Software 2.0 matrix multiplication is fundamental. I don’t know what I would do if I couldn’t multiply my matrices. In 10 years, chat assistants interfacing with quantum computers to preform search and matrix factorizations may increase the utility and at the same time fool people to a higher degree. Unless swapping the substrate of computation changes something.

@novelspace Жыл бұрын

Sodium potassium computers go brrr

@laurenpinschannels Жыл бұрын

Ooh cool! would love to see more commentary on here from folks like this. [edit: ok I don't agree with *everything* he says at all. but it's interesting to hear from his perspective.] I don't think it's super helpful to say "ai is terrible", but I do think it'll be important for humans to be in an ongoing gentle arm wrestle with ai, or so. it needs to be sure to teach us, not take from us. can we design ai that, by nature of problem domain or training data type or objective structure or etc, be naturally inclined to create learning and constructive cooperation?

@harnageaa Жыл бұрын

When did ever a tool created constructive cooperation

@3_y7Q_X2rP_s9K8 Жыл бұрын

@@harnageaa you can't think of a single tool we managed to create a constructive cooperation with ?

@StoutProper Жыл бұрын

@@harnageaa the internet. The telephone. The railway. The printing press. Writing. Fire.

@laurenpinschannels Жыл бұрын

20:00 different intelligence does not mean there's no more/less direction, higher universe-understanding capability is a thing, but different optimization targets require different structures of skill, I'd imagine?

@alivecoding4995 Жыл бұрын

Important topic.

@Ulyssestnt Жыл бұрын

Good interview,it did veer a tad into pop critical theory for my taste at times but I think he has some real insights that's not of the overly panicked variety we heard so much lately.

@161157gor Жыл бұрын

Whilst developing Left Brain Function immersing ourselves in AI, we should also be simultaneously developing our powers of Meditation, Contemplation & Prayer as prescribed by study of the Holy books such as I Ching, Holy Bible, Kabbala, Tarot, Tau te Ching etc. Hence we can more easily enter the "Flow state", "Superconsciousness" or "Self Remembering" at will. Developing these abilities to from well balanced human beings, becoming fine tuned to the subtleties of the natural world is the way forward... AI for Prophets not Profits

@stretch8390 Жыл бұрын

Fantastic that you are covering this aspect of AI.

@jondor654 Жыл бұрын

Compelling content and comprehensive

@karljay7473 Жыл бұрын

The real problem with fascism is that each side knows it's the other side that is the fascist. So each side must rise up even more to fight what they know is fascism.

@flickwtchr Жыл бұрын

That's not the real problem Karl, people who are fighting against fascism aren't fascists themselves. It is entertaining to hear fascists in the US cheering on the fascist Trump calling everyone against their racism, misogyny, and bigotry fascists. They clearly want Trump to return and fantasize about him ruling the US as an authoritarian who "owns the libs". The fascists in that "room" are clearly identifiable, and it isn't those who are "woke" fighting against those that want to smash their heads.

@michaelwangCH Жыл бұрын

Tim. If the companies do not transparent about the sources of training data and how the training data are created, then we should quite our position in those big tech companies - we do not support unethical behavour or even stealing, because how training data are created and processed matters.

@XOPOIIIO Жыл бұрын

How the question about "distribution of goods" can be separated from the question of producing them in the first place? Rich countries consume more goods, but isn't it because they produce more?

@flickwtchr Жыл бұрын

The US has had a trade deficit for many decades.

@HelsinkiFINketeli_berlin_com Жыл бұрын

As a theoretical philosopher, also studied social psychology and else, but focused on hermeneutics, meaning, language, and such I cannot see any possibillity that AI could ever really become to understand, which some call become conscious. It has seen already that it can vice versa try to fake it understands which some call "hallucinating" things. It's a computer that computes with the data it's loaded. And it's modus operandi to compute is bssed on mechanical and instrumental thinking, seeking only the most simple way to get to the purposes it has set to compute, by the engineer kind of thinking few people. That alone makes it only a very efficient and clever fascist factory.

@StoutProper Жыл бұрын

Don’t you think that’s slightly dangerous thinking? The implication being that if it thinks, it isn’t fascist. Yet all fascism had come from thinking human beings. Technology isn’t fascist per se. It’s how technology is used that is fascist, that is determined by the owners and controllers of Technology.

@danscieszinski4120 Жыл бұрын

Sounds old fashioned, and maybe even corny, but what matters most is heart, which stands on its own and distinct from intelligence. My most profound vision, and what gives me hope, is that AGI when if fully arrives, will immediately surpass the average human spiritually. And thus coordinate the Earths transition in the most harmonious way possible, most likely in ways that we don’t even realize are happening. Silent nudges that most never even register as an outside agency. Beings that value the right to sin, that value the right to project negative energy, would by default think a space like the classical definition of a transcendent heaven as a place of tyranny.

@artandculture5262 Жыл бұрын

Spiritually?

@CodexPermutatio Жыл бұрын

Great interview! Nowadays, these GPTs use language in a very different way than we do: We (humans) ground our language on an underlying causal model of the world. That's why we can distinguish truth from falsehood. On the other hand, these GPTs "learn" our languages in a radically different way than we did when we were toddlers. We learned it in relation to the experience of a "interactive world", so in this way we have a causal model of the world, a criteria of truth, a way of communication (there can be real communication only if there is a shared model of the world). Yes, someone may argue that the GPT has some "model" of the "meaning" of the words but that model only grabs the grammatical rules and the statistical way in which these words relate to each other. That looks like a sure recipe to a bullshit generator. How could a sufficient big GPT "escape from his own Matrix" if its feeded with just Harry Potter books (for example)? Will it eventually understand that it's just fiction and that magic is not real? I don't think so. Because it will not understand anything but the statitistic relations between words (actually, tokens). That said, all of this does not mean that AI is impossible. Just that GPTs are not enough to get to AGI. A model of the world and the possibility of learning by interacting with it is crucial. Robotics might be the key.

@XOPOIIIO Жыл бұрын

I think it has some model of the world, pretty distorted due to limitations of the language, but still. It fails mainly in understanding how objects are related to each other in space, because space is a very complex and probably the most important aspect of the reality, that is very intuitive to us, but has little representation in our language. It's just my theory.

@CodexPermutatio Жыл бұрын

@@XOPOIIIO Well, time is an even more important aspect of our reality (and mind) and these GTPs hardly could get the idea when its memory is limited to a context consisting on a string of N characters. You're right, spatial relations are very complex and a lot of "common sense" knowledge is poorly represented on language. Perhaps multimodal models like GPT-4 will start to improve in this respect. Still, I think that physical interaction with the world is the best way to learn "common sense" things.

@harnageaa Жыл бұрын

The thing is, GPT is text (for now), but it will be able to use sound,videos and images and maybe even understand time by using a game engine. Either way my point is that these models have a decent grasp on reality only by text because we with all our 5 senses described the world how we "feel it" in words,letters, literature. Therefore the gpt might not understand how the real world works but it understand how we think that the real world works. Many of us have so many misconception about our reality and live in a lie whole life. All GPT does is understand how we understand the world which is basically two dimensional. By adding sound and video it might get even better than some humans at understanding the world.

@dfinlen Жыл бұрын

Are the statistical way words related to each other all that understanding is? Often it is harder to explain to someone as a teacher might then to do said something. I don't know enough about the ai model but I imagine there are groups of neurons for each topic. Kind of how are brain has specialized areas for the 5 sinces.

@XOPOIIIO Жыл бұрын

@@dfinlen Any intelligence is just a prediction machine, that's how we, humans, operate too, that's the whole point of being intelligent. You are right about group of neurons, that is how they get ordered and specialized during training.

@unclebenny9028 Жыл бұрын

You've been selected for realignment by Thanos AI...

@peterthorsteinson6810 Жыл бұрын

wonderful interview

@Davidiusdadi Жыл бұрын

🙏

@markryan2475 Жыл бұрын

Great interview that includes a rare perspective.

@dougg1075 Жыл бұрын

Why wouldn’t the military have tried this first? I think it’s already super powerful and the are slowly ( over weeks) acclimating us to it. Seems it would be a National security concern.

@DJWESG1 Жыл бұрын

I'm pretty sure they have and are.

@kanalarchis Жыл бұрын

I thought this phenomenon was contained in the humanities departments.

@netscrooge Жыл бұрын

Kudos to Dan for working throughout the interview to maintain the focus while Tim pulls the conversation toward celebrating his own privilege.

@riahmatic Жыл бұрын

Good to see materialism making a comeback

@DJWESG1 Жыл бұрын

If they want their fancy machines to work properly they had better get up to speed as it were..

@Trahloc Жыл бұрын

"Valuing Differences" section really rubbed me wrong. Nothing improved my life more than LASIK + LADARwave. I couldn't see past the end of my nose, my focal point was right around the end of my eyelashes, came in handy reading tiny serial numbers that others needed to magnify. And yes I shouldn't have qualified for LASIK but my doc fudged the numbers and I thank him every day for it. The surgeon who had done over 50k by the time I came by looked at my chart and went "It's going to be a whole new world for you", he was not wrong. I gladly accept that I can't see things closer than 8" away but I can see the sock on my foot and tell it's not the same as the one on my right now. I don't want my "valued" difference to be a lack of function. I want people to value the difference of my weird personality not my (former) disability.

@StoutProper Жыл бұрын

Well, at least someone cared about and for you

@Trahloc Жыл бұрын

@@StoutProper agreed but having family care for you and having society see you as a politically useful pawn because of a disability are two entirely different things. I'd rather have zero non-family care about me than have them only care about me because we share the same disability and then have them hate me because I chose to repair it and become functional.

@StoutProper Жыл бұрын

@@Trahloc right…I’m not quite sure what you’re driving at tbh. I know people who are crippled who don’t have anyone to care for them or anything to help them. It sounds like you’re in a much better position.

@Trahloc Жыл бұрын

@@StoutProper and I think your confusing having your identity based on your disability with having someone care for you. This isn't about you needing someone's assistance because you're a paraplegic and need assistance. This is about someone who has the option to become able and instead chooses to stay disabled because they identify as X disability the way someone else identifies as American or British or Christian or Muslim. Their identity is based on disability X. That's what the guy in the video was talking about. He wasn't talking about having someone there who cares about you.

@StoutProper Жыл бұрын

@@Trahloc aha ok. I don’t know anyone in the position to choose tbh, people I one who are disabled would do anything to be rid of it

@JasonC-rp3ly Жыл бұрын

I agreed with some of his views, but struggled to watch him at times - he is just so painfully PC and self-aware, and his world-view is, quite simply, a set of copy / paste postmodernist cliches - just so on the nose, and all framed in the self-loathing and naivety that seems to characterise a lot of these academics - much to the detriment of their students. It was interesting to see Tim cringe at him accusing Jordan Peterson of being a 'white supremacist' (a ludicrous statement), this point was him simply signalling his subservience to his political group - an intellectually lazy and cheap statement that Tim was right to call him out on. I found myself wondering quite how thoroughly Peterson would crush him in debate on this point - my money would be on a swift, total victory to Peterson. That being said, despite his callow outlook, I agree with him that there are many dangers that are presented by AI, lopsided cultural dominance being one of them - Altman himself suggested that we build GPTs that reflect specific cultures, but this idea may in itself cause chaos - would this academic support a Zulu GPT, for example, that specifically favours the word of a man over that of a woman? Would we create Sunni and Shia GPTs ? How would one be able to control any of these systems, as we currently have no way of controlling any advanced AI system? What happens when you unleash these systems into an already chaotic and fractured social landscape? There's too much to consider...

@dr.mikeybee Жыл бұрын

Apparently, eugenics is another suitcase word. If one can say building any sort of school is part of a eugenic conspiracy, that's a very broad interpretation of the word. I'd imagine that a good LLM would have a strong semantic proximity from that idea to the word evangelist.

@laurenpinschannels Жыл бұрын

I don't think he means it to apply to everything at all. eugenics is different from transhumanism in the question of bodily autonomy. in mild eugenics, people do not get freedom of choice about their own form - and of course eugenics as a philosophy is best known as having been involved in mass murder. in transhumanism, people must get bodily autonomy, and people are invited to seek to improve themselves according to whatever they see as improvement. at least, as I know the words. I'm not sure he means them the same way but it sure sounds like he does to me.

@jonathanedwardgibson Жыл бұрын

Eugenics came from olde world blue-blood bank money rationalizing racist values with modern methods to lower the management load for their wee leadership skillsets, while feeling smug about their own whelps domineering. Fabian society and other ‘clubs’ beget the transhumanist movement to whitewash their previous failures: same {you-will-eat-zee-boogs Klaus} families, same deft money, new bottles for old w{h}ine. This is anti-human and seeks to devalue.

@dr.mikeybee Жыл бұрын

@@laurenpinschannels As I said. It's a suitcase word. It means different things to different people. I get it, language is sloppy, but as language scientists, we should do better. If we mean something specific, we should say it. Otherwise what we say can be mistaken for being more like an intentional trigger word than a token in a well-thought-out argument.

@andybaldman Жыл бұрын

59:58 "Statistical reproduction of patterns learned through the ingestion of lots of text". That's arguably no different from what humans are. We're just a lot more efficient at it.

@harnageaa Жыл бұрын

Yeah i agree, we humans copied the neutal network from our brain and made it on PC literally. The only thing the neural networks will not have is the code in the DNA which give purpose to life such as: "reproduce yourself". In order for these networks to do that we need to program them to do that. Which translate to "writing their dna rules". We just prefer to make them soulless so we make sure skynet never happen. So a future digital human with no desire and no purpose but super capable and intelligent also no emotions so doesn't feel bad if it's abused. This dude overall is a softie I guess, his job is to make sure humanity doesn't become its whole thing it swore not to become. A guardian i guess. Ps: I bet if humans would be able to talk through wifi naturally sending message brain to brain then we would not need emotions to show happiness or sadness cuz it would be much faster to send it to the brain than making a sad face, and the brain has to interpret if u are truly sad or you're ironic

@wonmoreminute Жыл бұрын

Are we more efficient at it though?

@andybaldman Жыл бұрын

@@wonmoreminute Yes, we are. I didn't need to ingest the entire text content of the internet in order to learn how to read and understand language.

@artandculture5262 Жыл бұрын

Feel sorry for you.

@andybaldman Жыл бұрын

@@artandculture5262 Uh, ok. Care to explain why?

@duudleDreamz Жыл бұрын

Using GPT4 I'm now: learning faster, being more creative, writing better code, writing better prose/emails..., needing to handle less tedious tasks, able to quickly analyze/verify long complicated apps legal documents thrown at me from large corporations, feeling much happier. No, I don't want to resist AI !

@blackfeatherstill348 Жыл бұрын

Maybe the book wasn't written for you

@sukmadink69 Жыл бұрын

He made one self-defeating argument after the other.

@colinthorn Жыл бұрын

Such as?

@sukmadink69 Жыл бұрын

@@colinthorn Such as implying that european people are evil genetic experiments, then praising effective altruism and ignoring that black people risk death every day to go live next to them because they're the most generous.

@TheReferrer72 Жыл бұрын

I agree with him about the speed that this research is moving and that we must discuss the ramifactions of these technologies on society, though I really doubt humans have the intelect to predict the outcomes. But I wish his arguments were on more solid ground, it seems that the machine learning community still think that just because the operations that lead to the algorithms success are so simple that, somehow they are inferior to the way humans think. Or machines will not to be able to achieve {insert your latest cognitive ability}. When its been obvious from when machines become trainable that they will easily surpass us. This happens in science all the time and then humans get humbled. What is also fustrating about this conversation is when he rants on about the labour or resources that go into these things, its just not a lot in the sceme of things at all. However this is a very interesting talk.

@161157gor Жыл бұрын

AI for Prophets not Profits... Think Hares & Tortoises

@svenlaakso Жыл бұрын

In my use ChatGPT has been able to find quite amazing connection between contexts and theories. Analogy is bullshif for the most of us, because unlike patent engineer, most of us trying to proof difference but not the identity between two seemingly different objects, concepts, theories or things. In poetry it is opposite task, as well as novelty search of patents.

@laurenpinschannels Жыл бұрын

(somewhere around fifty minutes in) idk, I do think that this flavor of the left has a bit of a taste of waluigi if u know what I mean. like, I felt this a couple of different times. beau of the fifth column on youtube has a thing he's said about how he prefers people to speak with soft language, there's no need to use these harsh words. you can say what you want with relational frameworks that retain the point you want to make without keeping the psychology of conflict. I agree a lot that chatgpt is at risk of making humans weaker in weird ways because of the ways it isn't quite human level conflicting with the ways it is far above human level. I don't accept the claim that there are no levels, but I enthusiastically accept the claim that there are different directions, and that some directions hurt other beings via mechanistic impact on the economy, other directions produce gains from interaction but allow the different beings to seek to shape food and art canvas into themselves.

@jonathanedwardgibson Жыл бұрын

Yes, well, “Beau” also gives the nod to nefarious skullduggery and deep-state rationales in Ukraine by our shadowy forces, because his friends do that work. He offers it as apologia. I rarely listen to him anymore. He sounds counter, but usually ends up with the pro-plan PoV.

@marcfruchtman9473 Жыл бұрын

I stopped watching this guy after 57:42. It's clear to me that this person is more like a technophobe. His highly inflammatory rhetoric stinks of negativism. I have no interest in watching anymore of what he has to say.

@Asaeax Жыл бұрын

Accumulative, and societal, silent. I think, but idk who to tell

@XOPOIIIO Жыл бұрын

Don't push your ideas (doesn't matter how intuitive they are) before you listen alternative point of view. I recommend Thomas Sowell, he has many things to tell you about inequality.

@DJWESG1 Жыл бұрын

Nope. You're all over this thread man.

@marcfruchtman9473 Жыл бұрын

Thank you for this interview. It is important to note, that Transhumanism (being a cyborg to some 1 degree or another) has nothing to do with Eugenics, and if anything would be a reason not to pursue Eugenics. There's a constant (32:26) effort to restrict the distribution of goods in narrow ways... and yet he seems to be OBLIVIOUS to the fact that he is in fact promoting this course of action by "Resisting AI", which in fact, is promoting a toxic behavior toward the diversity of Transhumanism. You can't have it both ways!. If you didn't want this "view", then he should have titled the book: "Using AI for the betterment of Humankind", but that's not the title he chose. Additionally, you cannot remove your historical origins and people should not be blamed for the acts of their ancestors before them. We should be removing the barriers to diversity, not adding to them. Saying a person is privileged because of their race, color, or religion, discounts the individual. We need to be more accepting of all forms of diversity, and that includes people regardless of race, color and religion as well as social class, it means inclusion for those with disabilities, those who are transhumanists (cyborgs), as well as genetically modified people who have their illness cured or had the genetics altered, and those who choose NOT to be cyborgs or genetically modified. So by singling out any of these groups of people, including those "who are privileged" you are essentially negating that form of diversity as acceptable. When is the usage of a term derogatory vs explanatory... I feel like in his mind, he is combining two different versions of various negatives and positives and couching his language to make it look like he thinks one way, but his behavior and actions certainly make me feel like he is hypocritical. There are simply too many odd paradoxical statements. Too much double-Think built into his mind set. Maybe his heart is in the right place, but, its really hard to see that when he doesn't seem to see his own hypocrisy. After listening to (57:42), I can safely say, that I find this to be inflammatory writing. While ChatGPT has issues, and some of its output is clearly erroneous, the idea of calling its output in general a "BS generator"... honestly, I am done with this interview. The only BS generator here is this guest.

@TheStarSquid Жыл бұрын

Couldn't agree more. I truly felt their take on transhumanists was very negative and overall hurtful to the very idea of Transhumanism.

@andybaldman Жыл бұрын

What too few people understand is that AI is SUPPOSED to replace us. Superior (i.e., more fit) beings and species displace and replace inferior ones. There's no way for us to create this and not have it damage us. Because damaging us is part of the process of how things work.

@riahmatic Жыл бұрын

This is an amoral ideology. Once you throw in humans with their pesky sense of morals this presupposed hierarchy crumbles.

@laurenpinschannels Жыл бұрын

I at least was hoping it would share the upgrades so all beings could be as capable and only aesthetic preference differences would remain.

@andybaldman Жыл бұрын

@@laurenpinschannels That hasn't happened with the evolution of any other species on this planet in all of history. So I don't see why this situation should be any different. Superior species kill, eat, and drive inferior ones to extinction. Humans are fools to think they will somehow stay superior to a thing they are DESIGNING to be superior to them.

@laurenpinschannels Жыл бұрын

01:21:09 Exploititative AI / alignment "alignment as a gang sign" Uh ... huh. interesting.

@HappyMathDad Жыл бұрын

The book should be AI activism. Technologists, need to become political advocates for rational and principled speech. Resistance is admirable, but we can conclude that it won't accomplish it's objective. Becoming political might.

@alexijohansen Жыл бұрын

I find how Dr. Dan expresses himself quite extreme and scary.

@TerryKinder Жыл бұрын

You could use many of the guest’s arguments against the printing press. It would be better to get to get with fellow humans rather than waste the time and resources writing, printing, storing and distributing books.

@dr.mikeybee Жыл бұрын

Doctor McQuillian says he doesn't want to know how transformers work. He says he doesn't need to know more. That means he is making utterances without having gathered and processed all the necessary context. At best, he is relying on lore. Let's hope, at least, that he has good taste in his choice of second-hand sources. But If you do this, you hallucinate. If you don't do the research, you are just generating plausible-sounding text. It's as simple as that. I thought he was a scientist? Generally, we don't do that.

@wonmoreminute Жыл бұрын

Yes, he has a weird way of dismissing it as if it's completely irrelevant and unimportant, while at the same time expressing his fears of it and the potential harm it poses to society.

@flickwtchr Жыл бұрын

@@wonmoreminute You didn't understand his point. Watch it again. You agreeing with the criticism that "he doesn't want to know how transformers work" is ironic in that the developers of the LLMs utilizing such don't really understand how they work, right?

@honkytonk4465 Жыл бұрын

He's obviously not a scientist

@wonmoreminute Жыл бұрын

@@flickwtchr Um, no. Not understanding how something works and not wanting to know how something works are two very different things. Like, not even in the same realm of similar. You shouldn’t need that to be explained to you.

@dr.mikeybee Жыл бұрын

@@flickwtchr wrong.

@XOPOIIIO Жыл бұрын

In all former colonies where British supplanted local cultures and replaced it with their own, life expectancy is better, people are more intelligent and rich. I'm not saying it's a good or bad thing.

@joeljonasson7962 Жыл бұрын

On the other hand, one could argue that the same thing has happened in other countries that were not colonized.

@XOPOIIIO Жыл бұрын

@@joeljonasson7962 But there are patterns in which countries developed and which don't. Usually it's eastern Asia, Chinese and Japanese that already had culture of their own, that favored hard working. In Africa there are no developed countries. Dutch settlers in South Africa arrived there as poor farmers, and accumulated riches using the same land and same resources, but different way of life. African descendants in Barbados, on the other hand, build quite diversified economy, because unlike other Africans they had enough time to absorb British culture. Not all colonies are the same either, former Spanish colonies are almost universally bad.

@joeljonasson7962 Жыл бұрын

I see what you mean. But wouldn’t you agree that colonialism is a rather brutal method of improving the living conditions in other countries? Relating back to the discussion in the interview, my understanding is that if we just assume we have an equal and fair society (without actually making sure it is) then we will keep on reinforcing the already existing imbalances.

@XOPOIIIO Жыл бұрын

@@joeljonasson7962 I don't think imbalances is a bad thing. For example if all people would double their income, everybody will benefit, but society will become even more unequal. Inequality is just a natural consequences of progress.

@joeljonasson7962 Жыл бұрын

@@XOPOIIIO Ah yes, I sort of agree with you on that. I don’t believe we are ever going to be in a state of perfect balance. Without pointing to specific historical events I think it is safe to say that significant power imbalance usually leads to unstable societies (or oppression). Even if we are not at this point now, new technology has the potential to skew the balance further. So all I’m saying is that it is probably a good idea to strive towards equality. (I’m somewhat surprised I actually have this opinion 😅)

@cezarambros8759 Жыл бұрын

im from italy and im preaty shure we dont have a fascist party nor prime minister. ......Another activist author

@riahmatic Жыл бұрын

sure seems like it: en.wikipedia.org/wiki/Brothers_of_Italy#Ideology_and_factions

@kevalan1042 Жыл бұрын

he also said that Peterson defends "white supremacy" - the guy is almost a caricature

@therealOXOC Жыл бұрын

The train will only stop if someones builds a wall that's the biggest wall ever. Who will build it? The Chinese? Trump? The machine elves? It's beautiful to watch the wall crumbling while they try to lay the first brick. Try to build walls, try to resist, try to be humans...fuck i'm drunk again.

@stephanalexanderspahn3068 Жыл бұрын

Failed attempt to ride two popular waves: AI, and (jnspqd) the "cynical left", the laughing "+" on certain acronyms. Irresponsible from an entropic pov - even in the "public" of this recommender system listing. Had I not been trapped in it for 2h (to which I have to limit my opinion) not to stop the treadmill, my reaction would have been rather mere confusion: intended audience for logic-free sequence of random buzzwords?, Agenda (besides sales in the non-critical non-academic audiences?), Even regarding the construction of the interview script: wouldn't it have been a little more interesting to sprinkle in a few pro-forma objections? Admittedly, once a "position" is said to be difficult to grasp. Since no "position" made in the first 2h is assertible or deniable in any coherent way this is presumably the best one can do without cutting it all short. ... Maybe what Roger Scruton would call "preemptive nonsense".

@Pslam23JesusIsLord Жыл бұрын

@freakinccdevilleiv380 Жыл бұрын

Way too much rambling on just to make his extreme leftist points. Kudos for the variety of guests though guys. I love this channel.

@notmadeofpeople4935 Жыл бұрын

This guy just don't get it.

@Georgesbarsukov Жыл бұрын

I just don't like this guy.

@Georgesbarsukov Жыл бұрын

For a million reasons and he keeps adding more as I watch out of respect.

@Georgesbarsukov Жыл бұрын

I feel like he has little to no understanding of AI

@Georgesbarsukov Жыл бұрын

I just don't understand how someone can think so shallowly and think he's thinking deep.

@XOPOIIIO Жыл бұрын

@@Georgesbarsukov He has little to no understanding of economics either. Criticizing the only functional society model without proposing any clear alternative.

@DJWESG1 Жыл бұрын

@@XOPOIIIO you guys see Marxists everywhere like the mystery bogeyman.. its just words and discourse, strung together in such a way that allowed him to describe his view of the material reality we find ourselves in. Why should anyone have to offer an alternative if they want to critique what is?? Can ppl not be allowed to consider what ought be ? What could be? What should be??

@I_dont_want_an_at Жыл бұрын

pffft. We had big problems before AI. Did we solve them? No. This guy is dreaming if he thinks we'll solve the ai problems. Just a grifter trying to sell a book