422 Matching Annotations
  1. Nov 2024
    1. the notes you make about it as you curate it (which tells the AI exactly what you find useful about it)

      My notes may give some indication about what I find useful about a thing, but certainly not exactly or even all of what I find useful.

    2. the model collapse paper now suggests that the training data created by well-managed communities could be the new currency of collective intelligence.
    3. “after greed and short-sightedness floods the commons with low-grade AI content… well-managed online communities of actual human beings [may be] the only place able to provide the sort of data tomorrow’s LLMs will need”

      The value spoken of here is that of slowly building up (evolving) directed knowledge over time. The community evolves links using work and coherence into actionable information/knowledge whereas AI currently don't have an idea of leadership or direction into which to take that knowledge, so they're just creating more related information which is interpreted as "adjacent noise". Choosing a path and building off of it to create direction is where the promise lies. Of course some paths may wither and die, but the community will manage that whereas the AI would indiscriminately keep building in all directions without the value of demise within the system.

  2. Oct 2024
    1. Successful Secretary Presented by Royal Office Typewriters. A Thomas Craven Film Corporation Production, 1966. https://www.youtube.com/watch?v=If5b2FiDaLk.

      Script: Lee Thuna<br /> Educational Consultant: Catharine Stevens<br /> Assistant Director: Willis F. Briley<br /> Design: Francisco Reynders<br /> Director & Producer: Carl A. Carbone<br /> A Thomas Craven Film Corporation Production

      "Mother the mail"

      gendered subservience

      "coding boobytraps"


      "I think you'll like the half sheet better. It is faster." —Mr. Typewriter, timestamp

      A little bit of the tone of "HAL" from 2001: A Space Odyssey (1968). This is particularly suggestive as H.A.L. was a one letter increment from I.B.M. and the 1966 Royal 660 was designed to compete with IBM's Selectric

      This calm voice makes suggestions to a secretary while H.A.L. does it for a male astronaut (a heroic figure of the time period). Suddenly the populace feels the computer might be a bad actor.

      "We're living in an electric world, more speed and less effort."—Mr. Typewriter<br /> (techno-utopianism)

    1. One employee suggested that Adobe should come up with "a long-term communication and marketing plan outside of blog posts," and meet with the company's most prominent critics on YouTube and social media to "correct the misinformation head-on.""Watching the misinformation spread on social media like wildfire is really disheartening," this person wrote in Slack. "Still, a loud 'F Adobe' and 'Cancel Adobe' rhetoric is happening within the independent creator community that needs to be addressed."A third worker said the internal communication review process might be broken. "What are we doing meaningfully to prevent this or is this only acted on when called out?" the person wrote.
    2. It's unclear how this information is accessed, and whether the creators of the data can opt out or get paid.
    3. Adobe upset many artists and designers recently by implying it would use their content to train AI models. The company had to quell those concerns with a blog post denying this.But some Adobe employees are still not happy with the response, and they are calling for improved communication with customers.According to screenshots of an internal Slack channel, obtained by Business Insider, Adobe employees complained about the company's poor response to the controversy and demanded a better long-term communication plan. They pointed out that Adobe got embroiled in similar controversies in the past, adding the internal review process needed to be fixed. This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Have an account? Log in. "If our goal is truly to prioritize our users' best interests (which, to be honest, I sometimes question), it's astonishing how poor our communication can be," one of the people wrote in Slack. "The general perception is: Adobe is an evil company that will do whatever it takes to F its users.""Let's avoid becoming like IBM, which seems to be surviving primarily due to its entrenched market position and legacy systems," this Adobe employee added.
  3. Sep 2024
    1. For all audiences and in most content, use intelligent or intelligence to describe or talk about the benefits of AI.In UI, use intelligent technology to describe the underlying technology that powers AI features.

      I think this is a good example of a misleading marketing ploy that shouldn't exist in technical documentation.

    1. the average percentage of hallucinated packages is at least 5.2% for commercial models and 21.7% for open-source models, including a staggering 205,474 unique examples of hallucinated package names, further underscoring the severity and pervasiveness of this threat
  4. Aug 2024
    1. Freedom of Expression and Challenges Posed by Artificial Intelligence

      The first section concentrates on the building a conceptual framework which is based on the relationship between freedom of expression and democracy and the way in which the Internet and social media enhance the exercise of freedom of expression.

      The second section examines the challenges to freedom of expression in relation to the promotion of gender equality on social media which are imposed by AI content curation and moderation.

      In the final section, the two-tiered approach, i.e., policies and law, is proposed to deal with the problem.

    1. we are slower we are irrational we are imperfect we are drifting away we are forgetting stuff we are making mistakes but we are learning from our failures we get support from our from our friends from our from our colleagues and we are understanding and instead of just analyzing the world and this is giving us the ultimate cognitive Edge

      for - key insight - human vs artificial intelligence - humans will create the best ideas

      key insight - human vs artificial intelligence - humans will create the best ideas - why? - because we are - slower - imperfect - less rational - drifting away - forgetting - and we learn from the mistakes we make and from different perspectives shared with us

    2. human beings don't do that we understand that the chair is not a specifically shaped object but something you consider and once you understood that concept that principle you see chairs everywhere you can create completely new chairs

      for - comparison - human vs artificial intelligence

      question - comparison - human vs artificial intelligence - Can't an AI also consider things we sit on to then generalize their classifcation algorithm?

    3. the brain is Islam Islam is it is lousy and it is selfish and still it is working yeah look around you working brains wherever you look and the reason for this is that we totally think differently than any kind of digital and computer system you know of and many Engineers from the AI field haven't figured out that massive difference that massive difference yet

      for - comparison - brain vs machine intelligence

      comparison - brain vs machine intelligence - the brain is inferior to machine in many ways - many times slower - much less accurate - network of neurons is mostly isolated in its own local environment, not connected to a global network like the internet - Yet, it is able to perform extraordinary things in spite of that - It is able to create meaning out of sensory inputs - Can we really say that a machine can do this?

    4. this blue ball with three stumps a chair or this strange design object here because you can sit on it and what you see here is the difference the main difference between the computer world and the brainworld

      for - comparison - brain vs machine intelligence - comparison - human intelligence vs artificial intelligence

      comparison - human intelligence vs artificial intelligence - AI depends on feeding the AI system with huge datasets that it can - analyze and make correlations and - perform big data analysis - Humans don't operate the same way

    5. for - Henning Beck - neuroscientist - video - youtube - The Brain vs Artificial Intelligence

  5. Jul 2024
    1. Please sign these letters to legislators, telling them that misguided AI laws will hurt startups and small companies and discourage AI innovation and investment in California.AI offers tremendous benefits, but many fear AI and worry about potential harm and misuse. These are valid concerns for everyone, including legislators, but laws that promote safe and equitable AI should be fact-based, straightforward, and universally applied. Legislators in Sacramento are considering two proposals, AB 2930 and SB 1047, that would impose costly and unpredictable burdens on AI developers, including anticipating and preventing future harmful uses of AI. Though well-intended, these bills will dampen and inhibit innovation, permanently embed today’s AI leaders as innovation gatekeepers, and drive investment and talent to other states and countries.
  6. Jun 2024
    1. In this respect, we join Fitzpatrick (2011) in exploring “the extent to which the means of media production and distribution are undergoing a process of radical democratization in the Web 2.0 era, and a desire to test the limits of that democratization”

      Comment by chrisaldrich: Something about this is reminiscent of WordPress' mission to democratize publishing. We can also compare it to Facebook whose (stated) mission is to connect people, while it's actual mission is to make money by seemingly radicalizing people to the extremes of our political spectrum.

      This highlights the fact that while many may look at content moderation on platforms like Facebook as removing their voices or deplatforming them in the case of people like Donald J. Trump or Alex Jones as an anti-democratic move. In fact it is not. Because of Facebooks active move to accelerate extreme ideas by pushing them algorithmically, they are actively be un-democratic. Democratic behavior on Facebook would look like one voice, one account and reach only commensurate with that person's standing in real life. Instead, the algorithmic timeline gives far outsized influence and reach to some of the most extreme voices on the platform. This is patently un-democratic.

  7. illuminate.withgoogle.com illuminate.withgoogle.com
    1. https://illuminate.withgoogle.com/

      via

      Interesting experiment from Google that creates an NPR-like discussion about any academic paper.<br><br>It definitely suggests some cool possibilities for science communication. And the voices, pauses, and breaths really scream public radio. Listen to at least the first 30 seconds. pic.twitter.com/r4ScqenF1d

      — Ethan Mollick (@emollick) June 1, 2024
      <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

  8. May 2024
    1. Meanwhile, Amazon and other ebook retails are pushing full-steam ahead to promote AI-generated content at the expense of real authors and artists. Publisher who actually pay authors and artists and editors now have to compete with AI-generated material churned out in bulk and sold at 99 cents. And while it is easy to shrug this off if you are outside the industry and claim, "Well, the cream rises to the top," anyone that has been around the industry long enough knows that what rises to the top is what Amazon's algorithms push there. And the AI bots are much better at manipulating the algorithms that real people.

      Amazon care about money; they don't care about humans.

  9. Apr 2024
    1. Have you ever had a meaningful conversation with Siri or Alexa or Cortana? Of course not.

      That said, I have had some pretty amazing conversations with ChatGPT-4. I've found it to be useful, too, for brainstorming. In one recent case (which I blogged about on my personal blog), the AI helped me through figuring out a structural issue with my zettelkasten.

    2. rtificial intelligence is already everywhere
    1. We’re over a year into the AI gold rush now, and corporations using top AI services report unremarkable gains, AI salesmen have been asked to rein in their promises for fear of underdelivering on them, an anti-generative AI cultural backlash is growing, the first high-profile piece of AI-centered consumer hardware crashed and burned in its big debut, and a bombshell scientific paper recently cast serious doubt on AI developers’ ability to continue to dramatically improve their models’ performance. On top of all that, the industry says that it can no longer accurately measure how good those models even are. We just have to take the companies at their word when they inform us that they’ve “improved capabilities” of their systems.
    1. "The problem is that even the best AI software can only take a poor-quality image so far, and such programs tend to over sharpen certain lines, resulting in strange artifacts," Foley said. Foley suggested that Netflix should have "at the very least" clarified that images had been altered "to avoid this kind of backlash," noting that "any kind of manipulation of photos in a documentary is controversial because the whole point is to present things as they were." Hollywood's increasing use of AI has indeed been controversial, with screenwriters' unions opposing AI tools as "plagiarism machines" and artists stirring recent backlash over the "experimental" use of AI art in a horror film. Even using AI for a movie poster, as Civil War did, is enough to generate controversy, the Hollywood Reporter reported.
    1. despite Rus’s assurances, people do fear that robots and AI will steal their jobs, or at best demote them to underlings. To counter this dystopian idea, Rus contends that technology usually creates jobs: between 1980 and 2015, for example, computers eliminated 3.5 million jobs but generated 19 million new ones. Perhaps so, but were these good jobs? Quality should count as much as quantity. And for every inspiring new application of robotics in the book, Rus includes another idea that made me cringe. On one page, she describes a smart glove that could help an elderly stroke victim regain the use of her hands and write a birthday card to a grandchild. Wonderful. Then she talks about adapting the glove for children, to take control of their hands when learning to write and circumvent the hard work of mastering this skill. Dreadful. In spite of her warnings about the need to properly train robots and AI systems, she seemingly forgets that human muscles and nerves need training as well, and that failing and flailing are integral parts of learning. Short-circuiting that process has a cost. Such lessons apply to adults, too. As much as I’d love to turn all yard work over to feral Roombas and never rake leaves or shovel snow again, such chores do instill discipline and pride. They’re also forms of physical activity, something most of us need more of. Indeed, it would be one thing if people offloaded tedious chores to robots and spent their free time hiking mountains or running marathons—but lying on the couch eating potato chips seems likelier. We’ve all seen the human blobs of Wall-E.
    1. it follows that no purchasable articlecan supply our individual wants so far as a key to our stockof information is concerned. We shall always be mainly de-pendent in this direction upon our own efforts to meet ourown situation.

      I appreciate his emphasis on "always" here. Though given our current rise of artificial intelligence and ChatGPT, this is obviously a problem which people are attempting to overcome.

      Sadly, AI seem to be designed for the commercial masses in the same way that Google Search is (cross reference: https://hypothes.is/a/jx6MYvETEe6Ip2OnCJnJbg), so without a large enough model of your own interests, can AI solve your personal problems? And if this is the case, how much data will it really need? To solve this problem, you need your own storehouse of personally curated data to teach an AI. Even if you have such a store for an AI, will the AI still proceed in the direction you would in reality or will it represent some stochastic or random process from the point it leaves your personal data set?

      How do we get around the chicken-and-egg problem here? What else might the solution space look like outside of this sketch?

  10. Mar 2024
    1. We can't use algorithms to filter for quality because they're not designed to. They're designed to steer you towards whatever's most profitable for their creators.That puts the onus on us, as users, to filter out the noise and that is increasingly difficult.
    1. So AI could be a tool to help move people into expression, to move past creative blocks

      To what extent are we using AI in this way in ds106? That is, using it as a starting point to build on rather than an end product?

    1. "The Curse of Recursion: Training on Generated Data Makes Models Forget," a recent paper, goes beyond the ick factor of AI that is fed on botshit and delves into the mathematical consequences of AI coprophagia: https://arxiv.org/abs/2305.17493 Co-author Ross Anderson summarizes the finding neatly: "using model-generated content in training causes irreversible defects": https://www.lightbluetouchpaper.org/2023/06/06/will-gpt-models-choke-on-their-own-exhaust/ Which is all to say: even if you accept the mystical proposition that more training data "solves" the AI problems that constitute total unsuitability for high-value applications that justify the trillions in valuation analysts are touting, that training data is going to be ever more elusive.
    2. Botshit can be produced at a scale and velocity that beggars the imagination. Consider that Amazon has had to cap the number of self-published "books" an author can submit to a mere three books per day: https://www.theguardian.com/books/2023/sep/20/amazon-restricts-authors-from-self-publishing-more-than-three-books-a-day-after-ai-concerns
    3. For people inflating the current AI hype bubble, this idea that making the AI "more powerful" will correct its defects is key. Whenever an AI "hallucinates" in a way that seems to disqualify it from the high-value applications that justify the torrent of investment in the field, boosters say, "Sure, the AI isn't good enough…yet. But once we shovel an order of magnitude more training data into the hopper, we'll solve that, because (as everyone knows) making the computer 'more powerful' solves the AI problem"
    4. As the lawyers say, this "cites facts not in evidence." But let's stipulate that it's true for a moment. If all we need to make the AI better is more training data, is that something we can count on? Consider the problem of "botshit," Andre Spicer and co's very useful coinage describing "inaccurate or fabricated content" shat out at scale by AIs: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4678265 "Botshit" was coined last December, but the internet is already drowning in it. Desperate people, confronted with an economy modeled on a high-speed game of musical chairs in which the opportunities for a decent livelihood grow ever scarcer, are being scammed into generating mountains of botshit in the hopes of securing the elusive "passive income": https://pluralistic.net/2024/01/15/passive-income-brainworms/#four-hour-work-week
  11. Feb 2024
    1. https://chat.openai.com/g/g-z5XcnT7cQ-zettel-critique-assistant

      Zettel Critique Assistant<br /> By Florian Lengyel<br /> Critique Zettels following three rules: Zettels should have a single focus, WikiLinks indicate a shift in focus, Zettels should be written for your future self. The GPT will suggest how to split multi-focused notes into separate notes. Create structure note from a list of note titles and abstracts.

      ᔥ[[ZettelDistraction]] in Share with us what is happening in your ZK this week. February 20, 2024

    1. Despite the opportunities of AI-based technologies for teaching and learning, they have also ethical issues.

      Yes, I agree with this statement. Ethical issues range from academic integrity concerns to data privacy. AI technology based on algorithmic applications intentionally collects human data from its users and they do not specifically know what kind of data and what quantities of them are collected

    1. Joy, Bill. “Why the Future Doesn’t Need Us.” Wired, April 1, 2000. https://www.wired.com/2000/04/joy-2/.

      Annotation url: urn:x-pdf:753822a812c861180bef23232a806ec0

      Annotations: https://jonudell.info/h/facet/?user=chrisaldrich&url=urn%3Ax-pdf%3A753822a812c861180bef23232a806ec0&max=100&exactTagSearch=true&expanded=true

    2. The experiences of the atomic scientists clearly show the need to takepersonal responsibility, the danger that things will move too fast, andthe way in which a process can take on a life of its own. We can, as theydid, create insurmountable problems in almost no time flat. We mustdo more thinking up front if we are not to be similarly surprised andshocked by the consequences of our inventions.

      Bill Joy's mention that insurmountable problems can "take on a life of [their] own" is a spectacular reason for having a solid definition of what "life" is, so that we might have better means of subverting it in specific and potentially catastrophic situations.

    3. The GNR technologies do not divide clearly into commercial andmilitary uses; given their potential in the market, it’s hard to imaginepursuing them only in national laboratories. With their widespreadcommercial pursuit, enforcing relinquishment will require a verificationregime similar to that for biological weapons, but on an unprecedentedscale. This, inevitably, will raise tensions between our individual pri-vacy and desire for proprietary information, and the need for verifica-tion to protect us all. We will undoubtedly encounter strong resistanceto this loss of privacy and freedom of action.

      While Joy looks at the Biological and Chemical Weapons Conventions as well as nuclear nonproliferation ideas, the entirety of what he's looking at is also embedded in the idea of gun control in the United States as well. We could choose better, but we actively choose against our better interests.

      What role does toxic capitalism have in pushing us towards these antithetical goals? The gun industry and gun lobby have had tremendous interest on that front. Surely ChatGPT and other LLM and AI tools will begin pushing on the profitmaking levers shortly.

    4. Now, as then, we are creators of new technologies and stars of theimagined future, driven—this time by great financial rewards andglobal competition—despite the clear dangers, hardly evaluating whatit may be like to try to live in a world that is the realistic outcome ofwhat we are creating and imagining.
  12. Jan 2024
    1. How soon could such an intelligent robot be built? The coming ad-vances in computing power seem to make it possible by 2030.

      In 2000, Bill Joy predicted that advances in computing would allow an intelligent robot to be built by 2030.

    2. in hishistory of such ideas, Darwin Among the Machines, George Dysonwarns: “In the game of life and evolution there are three players at thetable: human beings, nature, and machines. I am firmly on the side ofnature. But nature, I suspect, is on the side of the machines.”
    3. Uncontrolledself-replication in these newer technologies runs a much greater risk: arisk of substantial damage in the physical world.

      As a case in point, the self-replication of misinformation on social media networks has become a substantial physical risk in the early 21st century causing not only swings in elections, but riots, take overs, swings in the stock market (GameStop short squeeze January 2021), and mob killings. It is incredibly difficult to create risk assessments for these sorts of future harms.

      In biology, we see major damage to a wide variety of species as the result of uncontrolled self-replication. We call it cancer.

      We also see programmed processes in biological settings including apoptosis and necrosis as means of avoiding major harms. What might these look like with respect to artificial intelligence?

    4. Moravec’s view is that the robots will eventually suc-ceed us—that humans clearly face extinction.

      Joy contends that one of Hans Moravec's views in his book Robot: Mere Machine to Transcendent Mind is that robots will push the human species into extinction in much the same way that early North American placental species eliminated the South American marsupials.

    5. Our overuse of antibiotics has led to what may be thebiggest such problem so far: the emergence of antibiotic-resistant andmuch more dangerous bacteria. Similar things happened when attemptsto eliminate malarial mosquitoes using DDT caused them to acquireDDT resistance; malarial parasites likewise acquired multi-drug-resistant genes.

      Just as mosquitoes can "acquire" (evolve) DDT resistance or bacteria might evolve antiobiotic-resistance, might not humans evolve AI resistance? How fast might we do this? On what timeline? Will the pressure be slowly built up over time, or will the onset be so quick that extinction is the only outcome?

    1. Use Glaze, a system designed to protect human artists by disrupting style mimicry, to protect what you create from being stolen under the guise of 'training AI'; the term should really be 'thievery'.

  13. Dec 2023
    1. Matt GrossMatt Gross (He/Him) • 1st (He/Him) • 1st Vice President, Digital Initiatives at Archetype MediaVice President, Digital Initiatives at Archetype Media 4d • 4d • So, here's an interesting project I launched two weeks ago: The HistoryNet Podcast, a mostly automated transformation of HistoryNet's archive of 25,000+ stories into an AI-driven daily podcast, powered by Instaread and Zapier. The voices are pretty good! The stories are better than pretty good! The implications are... maybe terrifying? Curious to hear what you think. Listen at https://lnkd.in/emUTduyC or, as they always say, "wherever you get your podcasts."

      https://www.linkedin.com/feed/update/urn:li:activity:7142905086325780480/

      One can now relatively easily use various tools in combination with artificial intelligence-based voices and reading to convert large corpuses of text into audiobooks, podcasts or other spoken media.

    1. there's this broader issue of of being able to get inside other people's heads as we're driving down the road all the time we're looking at other 00:48:05 people and because we have very advanced theories of mind
      • for: comparison - AI - HI - example - driving, comparison - artificial i human intelligence - example - driving
    2. in my view the biggest the most dangerous phenomenon on the human on our planet is uh human stupidity it's not artificial intelligence
      • for: meme - human stupidity is more dangerous than artificial intelligence

      • meme: human stupidity is more dangerous than artificial intelligence

      • author:Nikola Danaylov
      • date: 2021
  14. Nov 2023
    1. I use expiration dates and refrigerators to make a point about #AI and over-reliance, and @dajb uses ducks. #nailingit @weareopencoop

      —epilepticrabbit @epilepticrabbit@social.coop on Nov 09, 2023, 11:51 at https://mastodon.social/@epilepticrabbit@social.coop/111382329524902140

    1. As an ex-Viv (w/ Siri team) eng, let me help ease everyone's future trauma as well with the Fundamentals of Assisted Intelligence.<br><br>Make no mistake, OpenAI is building a new kind of computer, beyond just an LLM for a middleware / frontend. Key parts they'll need to pull it off:… https://t.co/uIbMChqRF9

      — Rob Phillips 🤖🦾 (@iwasrobbed) October 29, 2023
      <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
  15. Oct 2023
    1. Wang et. al. "Scientific discovery in the age of artificial intelligence", Nature, 2023.

      A paper about the current state of using AI/ML for scientific discovery, connected with the AI4Science workshops at major conferences.

      (NOTE: since Springer/Nature don't allow public pdfs to be linked without a paywall, we can't use hypothesis directly on the pdf of the paper, this link is to the website version of it which is what we'll use to guide discussion during the reading group.)

    1. Envisioning the next wave of emergent AI

      Are we stretching too far by saying that AI are currently emergent? Isn't this like saying that card indexes of the early 20th century are computers. In reality they were data storage and the "computing" took place when humans did the actual data processing/thinking to come up with new results.

      Emergence would seem to actually be the point which comes about when the AI takes its own output and continues processing (successfully) on it.

  16. Sep 2023
    1. R.U.R.: Rossum’s Universal Robots, drama in three acts by Karel Čapek, published in 1920 and performed in 1921. This cautionary play, for which Čapek invented the word robot (derived from the Czech word for forced labour), involves a scientist named Rossum who discovers the secret of creating humanlike machines. He establishes a factory to produce and distribute these mechanisms worldwide. Another scientist decides to make the robots more human, which he does by gradually adding such traits as the capacity to feel pain. Years later, the robots, who were created to serve humans, have come to dominate them completely.
    1. What do you do then? You can take the book to someone else who, you think, can read better than you, and have him explain the parts that trouble you. ("He" may be a living person or another book-a commentary or textbook. )

      This may be an interesting use case for artificial intelligence tools like ChatGPT which can provide the reader of complex material with simplified synopses to allow better penetration of the material (potentially by removing jargon, argot, etc.)

    2. Active Reading

      He then pushes a button and "plays back" the opinion whenever it seems appropriate to do so. He has performed acceptably without having had to think.

      This seems to be a reasonable argument to make for those who ask, why read? why take notes? especially when we can use search and artificial intelligence to do the work for us. Can we really?

  17. Aug 2023