KYield, Inc.’s Post

A thoughtful discussion with Shannon Vallor, written by Philip Ball, on the current manifestation and hype-storm of AI, and the extreme elements in SV that have evolved into what is undeniably a religious cult in LLM firms and portions of Big Techs. I'm in substantial agreement with Shannon on most points. The importance of AI for good can't be dismissed, for example, and no question about the recent trend towards authoritarianism -- even before LLMs, though misused AI is a near perfect tool for them (e.g., the CCP). However, it's important to understand and highlight the difference between Hinton's fear of rogue super intelligence that decides to destroy humanity, which I've always said was far too premature and distracting, and the greatest risks consumer LLM chatbots have already created. LLMs are indeed stochastic parrots, but they are more than that. Due to the vast, unprecedented data scale scraped on every topic, including scientific journals, run on the most powerful supercomputers, combined with inherent security flaws that invite jailbreaks by evildoers, as I posted earlier today -- LLMs are among the top two or three cat risks today. When combined with nuclear bombs, bioweapons/pandemic risk, and efforts to undermine civilization -- particularly by authoritarian regimes, I think consumer LLMs made available to the public represent the greatest risk to our species today. Moreover, these types of risks evolve quickly in chain reactions, undoubtedly to include 'sleeper cells' in state-backed groups. Melanie Mitchell and David Krakauer are probably correct in suggesting that it may represent a new type of intelligence. At very least it's the largest and most dynamic representation of collective intelligence to have ever been unleashed, for better and worse, with all the limitations and risks inherent with the technology, the way it's been recklessly applied, and the powerful perverse incentives driving it.

AI and the Battle for Truth: Between the Peril and the Promise. The battle against “fake news” has shaken the foundations of democracy, but with artificial intelligence (AI) on the rise, this may only be the beginning. Disinformation threatens to spread like a virus. Is there a way to harness this technology without devastating democracy? A just-released federal government study, Disruptions on the Horizon, is not encouraging. It identifies 35 plausible future threats to Canada and the world. Topping the list is the inability to distinguish truth from falsehood, with rogue AI not far behind. But as one AI expert notes, "We're always very careful [when] we talk about AI: It's the promise and the peril, they come hand in hand." So, what can be done to realize the promise and avoid the peril? Let’s dig a little deeper. https://lnkd.in/eKjkpDCf

🔎 Discover Transcrime's research in the #FermiEU project, presented at #ASC2024 by Michael Victor Lo Giudice! It analyses the nexus of racially and politically charged #conspiracytheories driving real-world #hatecrime from an undervalued angle: rather than focusing on the supply of harmful content (articles, blogs and posts), the spotlight was turned at its demand. 👉 Find out how the predictive power of AI has been used in #FermiEU to assess the impact of the spread of #disinformation online: https://lnkd.in/edNM62Ar Authors: Michael Victor Lo Giudice, Alberto Aziani, Ali Shadman

Heya! I have been formalizing my thoughts over the past few weeks about the current state of AI, how we're already in the intelligence era, and what the future looks like, over a broad spectrum of fields, from generative AI, AI agents, to software development, to jobs, to legal, ethical and moral ramifications, to the impending energy crunch we'll find ourselves in, in the near future. Give it a read! https://lnkd.in/gxVFQuSs

My technology turns an LLM model into a Hive Mind. The LLM Model becomes like the Queen. She instructs the workers, she oversees the workers, she has two way communication with the workers. The workers have a bit of their own intelligence built in. They function as an entity on their own, but instructions from the Queen are the highest priority in their hierarchy. If they Queen instructs them to jump off a bridge for the good of the hive, the individual worker jumps off a bridge. None of this is new on its own. Particle Swarm Optimization was invented in 1995. I remember being fascinated by it as a child. LLM models are LLM models. I make a few novel adjustments to the shape and function of these two elements to fuse them, but that is about it. They simply go together. People always say, "don't anthropomorphize AI". So, don't anthropomorphize AI intelligence either. It has all of the characteristics of a hive mind.

For anyone interested in investigating the potential for legitimate interest as the legal basis for AI systems….look no further. This year long awaited report is here from the Information Accountability Foundation. This will be very helpful for all DPOs. Thanks to Luis Alberto Montezuma for sharing. #AIGovernance #dataprotection #privacy #leadership #sheleadstech

AI and LLM research have been stuck for 7 years. Brute force dominates, rejecting principled understanding. The industry must admit the lack of true intelligence in current systems to make real progress and stop optimizing dead ends.

Are we witnessing the dawn of machines that truly think? Or are we merely creating increasingly convincing illusions of #intelligence? These questions lie at the heart of a heated debate currently dividing the #AI community. On one side, there is OpenAI's latest breakthrough with their "o1" model, promising #reasoning capabilities closer to human #intelligence, showing remarkable performance in complex tasks, particularly in science, coding, and math, outperforming its predecessors by a significant margin. As Sam Altman recently said, OpenAI's achievements could suggest we're on the cusp of Artificial General Intelligence (in a recent interview he said AGI will be a reality in "5 years, give or take, maybe slightly longer") (link in the comment). On the other side stands Yann LeCun, Meta's chief AI scientist, is offering a sobering counterpoint. LeCun argues that current AI models, despite their impressive abilities, are far from rivaling even animal intelligence. He dismisses warnings about AI's existential threats as "complete B.S.," challenging the narrative pushed by some of his peers and tech luminaries. The crux of this debate lies in the nature of #intelligence itself. OpenAI's "o1" model employs a "chain of thought" technique, allowing it to recognize and correct mistakes, break down complex problems, and try alternative approaches. This mimics human problem-solving strategies more closely than ever before, seemingly bringing us tantalizingly close to AGI. However, LeCun contends that true intelligence requires fundamental components absent in current systems: a mental model of the physical world, persistent #memory, reasoning ability, and capacity for planning. He argues that while large language models excel at manipulating language, this doesn't equate to genuine understanding or intelligence. Fascinating debate, right? It's a debate which shows striking similarities to other pivotal moments in the History of Science. One parallel is the development of quantum mechanics in the early 20th century. Like today's AI researchers, physicists were grappling with a paradigm shift that challenged fundamental assumptions about reality. The Copenhagen interpretation faced opposition from skeptics like #Einstein, much as LeCun challenges the optimistic narratives surrounding current AI progress. Another historical parallel can be drawn with the development of evolutionary theory. #Darwin's theory of natural selection faced significant opposition from those who saw it as a challenge to established beliefs. It's a debate which reminds us that #Progress (with capital P) often comes through vigorous debate and the challenging of established #paradigms. As we move forward, it's essential to appreciate the progress that have been made while maintaining a critical and open-minded approach to the gigantic challenge of creating truly intelligent machines. #NewFuture

The upshot of this is exactly the argument I made in my blog post this week regarding the need to introduce governance for understanding the translation between human requirements and what an AI can compute, and the very real risk of automation bias, seemingly taken in this case to excess (https://lnkd.in/etkanNXB). It also seems clear that while the inputs taken to inform the model (diagnosis, age, living situation) can reflect on average what a patient’s needs in similar circumstances may be, it cannot predict the exact scenario of any one patient, nor can it determine whether a claim should be approved. The case also stresses that AIs should be understood as parts of a system as well as models in their own right. In this instance, the algorithm itself did not have a 90% error rate (contrary to reports and the lawsuit itself which states the decisions were, "based on an AI model that Defendants know has a 90% error rate.") However, the system implementing the algorithm (humans and AI in combination) seems to have been extremely flawed, leading to a huge number of wrongful denials. My concern is that we will see increasing instances of cases such as this. It is tragic that it has taken a murder for a very public light to be shone on this. The very real risk of such purported incorrect and unethical use of AI is a public backlash against AI. This happened with genetically modified crops in Europe in the 1990s and aspects of nanotechnology in the 2000s. The pronouncements of AI evangelists (in which I include myself) will be seen as self-serving hyperbole if they are not combined with necessary and appropriate guardrails. The result risks putting AI research back by decades and/or leading to punitive legislation which will make the EU's AI Act seem tame. Please note that this article has been carefully worded. It reflects my reading of the referenced lawsuit and articles and is not the result of original research. It is also my own work and views and does not reflect the views or opinions of my employer. Post 3 of 3 Full post at https://lnkd.in/eVXcFED5

Artificial intelligence (AI) has been billed as the next frontier of humanity: the newly available expanse whose exploration will drive the next era of growth, wealth, and human flourishing. It’s a scary metaphor. Throughout American history, the drive for expansion and the very concept of terrain up for grabs—land grabs, gold rushes, new frontiers—have provided a permission structure for imperialism and exploitation. This could easily hold true for AI. https://lnkd.in/eqj4bDHq