195 Matching Annotations
  1. Aug 2022
    1. I recall being told by a distinguishedanthropological linguist, in 1953, that he had no intention of working througha vast collection of materials that he had assembled because within a few yearsit would surely be possible to program a computer to construct a grammar froma large corpus of data by the use of techniques that were already fairly wellformalized.

      rose colored glasses...

    1. For the sake of simplicity, go to Graph Analysis Settings and disable everything but Co-Citations, Jaccard, Adamic Adar, and Label Propogation. I won't spend my time explaining each because you can find those in the net, but these are essentially algorithms that find connections for you. Co-Citations, for example, uses second order links or links of links, which could generate ideas or help you create indexes. It essentially automates looking through the backlinks and local graphs as it generates possible relations for you.
  2. Jul 2022
    1. AI text generator, a boon for bloggers? A test report

      While I wanted to investigate AI text generators further, I ended up writing a testreport.. I was quite stunned because the AI ​​text generator turns out to be able to create a fully cohesive and to-the-point article in minutes. Here is the test report.

    1. A cognitiveagent is needed to perform this very action (that needs to be recurrent)—and another agent is neededto further build on that (again recurrently and irrespective to the particular agents involved).

      This appears to be setting up the conditions for an artificial cognitive agent to be able to play a role (ie Artificial Intelligence)

    2. In this paper, we propose and analyse a potential power triangle between three kinds of mutuallydependent, mutually threatening and co-evolving cognitive systems—the human being, the socialsystem and the emerging synthetic intelligence. The question we address is what configuration betweenthese powers would enable humans to start governing the global socio-econo-political system
      • Optimization problem - human beings, their social system and AI - what is optimal configuration?
    1. Superintelligence has long served as a source of inspiration for dystopian science fiction that showed humanity being overthrown, defeated, or imprisoned by machines.
  3. Jun 2022
    1. We've yet to see note-taking platforms meaningfully add AI affordances into their systems, but there are hints at how they could in other platforms.

      A promising project is Paul Bricman's Conceptarium.

    1. Dall-E delivers ten images for each request, and when you see results that contain sensitive or biased content, you can flag them to OpenAI for review. The question then becomes whether OpenAI wants Dall-E's results to reflect society's approximate reality or some idealized version. If an occupation is majority male or female, for instance, and you ask Dall-E to illustrate someone doing that job, the results can either reflect the actual proportion in society, or some even split between genders. They can also account for race, weight, and other factors. So far, OpenAI is still researching how exactly to structure these results. But as it learns, it knows it has choices to make.

      Philosophical questions for AI-generated artwork

      As if we needed more technology to dissolve a shared, cohesive view of reality, we need to consider how it is possible to tune the AI parameters to reflect some version of what is versus some version of how we want it to be.

    1. Harness collective intelligence augmented by digital technology, and unlock exponential innovation. Beyond old hierarchical structures and archaic tools.

      https://twitter.com/augmented_CI

      The words "beyond", "hierarchical", and "archaic" are all designed to marginalize prior thought and tools which all work, and are likely upon which this broader idea is built. This is a potentially toxic means of creating "power over" this prior art rather than a more open spirit of "power with".

  4. May 2022
    1. Bret Victor shared this post to make the point that we shouldn't be worrying about sentient AI right now; that the melting ice caps are way more of a threat than AGI. He linked to this article, saying that corporations act like a non-human, intelligent entity, that has real impacts in the world today, that may be way more consequential than AI.

    1. Ben Williamson shared this post on Twitter, saying that it's a good idea to remove the words 'artificial intelligence' and 'AI' from policy statements, etc. as a way of talking about specific details of a technology. We can see loads of examples of companies using 'AI' to obfuscate what they are really going.

    1. The bulk of Vumacam’s subscribers have thus far been private security companies like AI Surveillance, which supply anything from armed guards to monitoring for a wide range of clients, including schools, businesses, and residential neighborhoods. This was always the plan: Vumacam CEO Croock started AI Surveillance with Nichol shortly after founding Vumacam and then stepped away to avoid conflicts with other Vumacam customers.

      AI-driven Surveillance-as-a-Service

      Vumacam provides the platform, AI-driven target selection, and human review. Others subscribe to that service and add their own layers of services to customers.

  5. Apr 2022
    1. Since most of our feeds rely on either machine algorithms or human curation, there is very little control over what we actually want to see.

      While algorithmic feeds and "artificial intelligences" might control large swaths of what we see in our passive acquisition modes, we can and certainly should spend more of our time in active search modes which don't employ these tools or methods.

      How might we better blend our passive and active modes of search and discovery while still having and maintaining the value of serendipity in our workflows?

      Consider the loss of library stacks in our research workflows? We've lost some of the serendipity of seeing the book titles on the shelf that are adjacent to the one we're looking for. What about the books just above and below it? How do we replicate that sort of serendipity into our digital world?

      How do we help prevent the shiny object syndrome? How can stay on task rather than move onto the next pretty thing or topic presented to us by an algorithmic feed so that we can accomplish the task we set out to do? Certainly bookmarking a thing or a topic for later follow up can be useful so we don't go too far afield, but what other methods might we use? How can we optimize our random walks through life and a sea of information to tie disparate parts of everything together? Do we need to only rely on doing it as a broader species? Can smaller subgroups accomplish this if carefully planned or is exploring the problem space only possible at mass scale? And even then we may be under shooting the goal by an order of magnitude (or ten)?

    1. ResearchRabbit, which fully launched in August 2021, describes itself as “Spotify for papers”.

      Research Rabbit is a search engine for academic research that was launched in August of 2021 and bills itself as "Spotify for papers." It uses artificial intelligence to recommend related papers to researchers and updates those recommendations based on the contents of one's growing corpus of interest.

    2. Connected Papers uses the publicly available corpus compiled by Semantic Scholar — a tool set up in 2015 by the Allen Institute for Artificial Intelligence in Seattle, Washington — amounting to around 200 million articles, including preprints.

      Semantic Scholar is a digital tool created by the Allen Institute for Artificial Intelligence in Seattle, Washington in 2015. It's corpus is publicly available for search and is used by other tools including Connected Papers.

    1. He continues by comparing open works to Quantum mechanics, and he arrives at the conclusion that open works are more like Einstein's idea of the universe, which is governed by precise laws but seems random at first. The artist in those open works arranges the work carefully so it could be re-organized by another but still keep the original voice or intent of the artist.

      Is physics open or closed?

      Could a play, made in a zettelkasten-like structure, be performed in a way so as to keep a consistent authorial voice?

      What potential applications does the idea of opera aperta have for artificial intelligence? Can it be created in such a way as to give an artificial brain a consistent "authorial voice"?

  6. Mar 2022
    1. projet européen X5-GON (Global Open Education Network) qui collecte les informations sur les ressources éducatives libres et qui marche bien avec un gros apport d’intelligence artificielle pour analyser en profondeur les documents
    1. This generative model normally penalizes predicted toxicity and rewards predicted target activity. We simply proposed to invert this logic by using the same approach to design molecules de novo, but now guiding the model to reward both toxicity and bioactivity instead.

      By changing the parameters of the AI, the output of the AI changed dramatically.

    1. Of course, users are still the source of the insight that makes a complete document also a compelling document.

      Nice that he takes a more humanistic viewpoint here rather than indicating that it will all be artificial intelligence in the future.

  7. Feb 2022
    1. Stay at the forefront of educational innovation

      What about a standard of care for students?

      Bragging about students not knowing how the surveillance technology works is unethical.<br><br>Students using accessibility software or open educational resources shouldn't be punished for accidentally avoiding surveillance. pic.twitter.com/Uv7fiAm0a3

      — Ian Linkletter (@Linkletter) February 22, 2022
      <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

      #annotation https://t.co/wVemEk2yao

      — Remi Kalir (@remikalir) February 23, 2022
      <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
    1. At the back of Dr Duncan's book on the topic, Index, A History Of The, he includes not one but two indexes, in order to make a point.

      Dennis Duncan includes two indices in his book Index, A History of The, one by a professional human indexer and the second generated by artificial intelligence. He indicates that the human version is far better.

    1. We need to getour thoughts on paper first and improve them there, where we canlook at them. Especially complex ideas are difficult to turn into alinear text in the head alone. If we try to please the critical readerinstantly, our workflow would come to a standstill. We tend to callextremely slow writers, who always try to write as if for print,perfectionists. Even though it sounds like praise for extremeprofessionalism, it is not: A real professional would wait until it wastime for proofreading, so he or she can focus on one thing at a time.While proofreading requires more focused attention, finding the rightwords during writing requires much more floating attention.

      Proofreading while rewriting, structuring, or doing the thinking or creative parts of writing is a form of bikeshedding. It is easy to focus on the small and picayune fixes when writing, but this distracts from the more important parts of the work which really need one's attention to be successful.

      Get your ideas down on paper and only afterwards work on proofreading at the end. Switching contexts from thinking and creativity to spelling, small bits of grammar, and typography can be taxing from the perspective of trying to multi-task.


      Link: Draft #4 and using Webster's 1913 dictionary for choosing better words/verbiage as a discrete step within the rewrite.


      Linked to above: Are there other dictionaries, thesauruses, books of quotations, or individual commonplace books, waste books that can serve as resources for finding better words, phrases, or phrasing when writing? Imagine searching through Thoreau's commonplace book for finding interesting turns of phrase. Naturally searching through one's own commonplace book is a great place to start, if you're saving those sorts of things, especially from fiction.

      Link this to Robin Sloan's AI talk and using artificial intelligence and corpuses of literature to generate writing.

  8. Jan 2022
    1. https://vimeo.com/232545219

      from: Eyeo Conference 2017

      Description

      Robin Sloan at Eyeo 2017 | Writing with the Machine | Language models built with recurrent neural networks are advancing the state of the art on what feels like a weekly basis; off-the-shelf code is capable of astonishing mimicry and composition. What happens, though, when we take those models off the command line and put them into an interactive writing environment? In this talk Robin presents demos of several tools, including one presented here for the first time. He discusses motivations and process, shares some technical tips, proposes a course for the future — and along the way, write at least one short story together with the audience: all of us, and the machine.

      Notes

      Robin created a corpus using If Magazine and Galaxy Magazine from the Internet Archive and used it as a writing tool. He talks about using a few other models for generating text.

      Some of the idea here is reminiscent of the way John McPhee used the 1913 Webster Dictionary for finding words (or le mot juste) for his work, as tangentially suggested in Draft #4 in The New Yorker (2013-04-22)

      Cross reference: https://hypothes.is/a/t2a9_pTQEeuNSDf16lq3qw and https://hypothes.is/a/vUG82pTOEeu6Z99lBsrRrg from https://jsomers.net/blog/dictionary


      Croatian acapella singing: klapa https://www.youtube.com/watch?v=sciwtWcfdH4


      Writing using the adjacent possible.


      Corpus building as an art [~37:00]

      Forgetting what one trained their model on and then seeing the unexpected come out of it. This is similar to Luhmann's use of the zettelkasten as a serendipitous writing partner.

      Open questions

      How might we use information theory to do this more easily?

      What does a person or machine's "hand" look like in the long term with these tools?

      Can we use corpus linguistics in reverse for this?

      What sources would you use to train your model?

      References:

      • Andrej Karpathy. 2015. "The Unreasonable Effectiveness of Recurrent Neural Networks"
      • Samuel R. Bowman, Luke Vilnis, Oriol Vinyals, et al. "Generating sentences from a continuous space." 2015. arXiv: 1511.06349
      • Stanislau Semeniuta, Aliaksei Severyn, and Erhardt Barth. 2017. "A Hybrid Convolutional Variational Autoencoder for Text generation." arXiv:1702.02390
      • Soroush Mehri, et al. 2017. "SampleRNN: An Unconditional End-to-End Neural Audio Generation Model." arXiv:1612.07837 applies neural networks to sound and sound production
    1. Markoff, a long-time chronicler of computing, sees Engelbart as one pole in a decades-long competition "between artificial intelligence and intelligence augmentation -- A.I. versus I.A."

      There is an interesting difference between artificial intelligence and intelligence automation. Index cards were already doing the second by the early 1940s.

  9. Dec 2021
  10. Nov 2021
  11. Oct 2021
  12. Sep 2021
  13. Aug 2021
    1. Provide more opportunities for new talent. Because healthcare has been relatively solid and stagnant in what it does, we're losing out on some of the new talent that comes out — who are developing artificial intelligence, who are working at high-tech firms — and those firms can pay significantly higher than hospitals for those talents. We have to find a way to provide some opportunities for that and apply those technologies to make improvements in healthcare.

      Intestesing. Mr. Roach thinks healthcare is not doing enough to attract new types of talent (AI and emerging tech) into healthcare. We seem to be losing this talent to the technology sector.

      I would agree with this point. Why work for healthcare with all of its massive demands and HIPPA and lack of people knowing what you are even building. Instead, you can go into tech, have a better quality of life, get paid so much more, and have the possibility of exiting due to a buyout from the healthcare industry.

    1. Building on platforms' stores of user-generated content, competing middleware services could offer feeds curated according to alternate ranking, labeling, or content-moderation rules.

      Already I can see too many companies relying on artificial intelligence to sort and filter this material and it has the ability to cause even worse nth degree level problems.

      Allowing the end user to easily control the content curation and filtering will be absolutely necessary, and even then, customer desire to do this will likely loose out to the automaticity of AI. Customer laziness will likely win the day on this, so the design around it must be robust.

  14. Jul 2021
    1. Facebook AI. (2021, July 16). We’ve built and open-sourced BlenderBot 2.0, the first #chatbot that can store and access long-term memory, search the internet for timely information, and converse intelligently on nearly any topic. It’s a significant advancement in conversational AI. https://t.co/H17Dk6m1Vx https://t.co/0BC5oQMEck [Tweet]. @facebookai. https://twitter.com/facebookai/status/1416029884179271684

  15. Jun 2021
    1. t hadn’t learned sort of the concept of a paddle or the concept of a ball. It only learned about patterns of pixels.

      Cognition and perception are closely related in humans, as the theory of embodied cognition has shown. But until the concept of embodied cognition gained traction, we had developed a pretty intellectual concept of cognition: as something located in our brains, drained of emotions, utterly rational, deterministic, logical, and so on. This is still the concept of intelligence that rules research in AI.

    2. the original goal at least, was to have a machine that could be like a human, in that the machine could do many tasks and could learn something in one domain, like if I learned how to play checkers maybe that would help me learn better how to play chess or other similar games, or even that I could use things that I’d learned in chess in other areas of life, that we sort of have this ability to generalize the things that we know or the things that we’ve learned and apply it to many different kinds of situations. But this is something that’s eluded AI systems for its entire history.

      The truth is we do not need to have computers to excel in the things we do best, but to complement us. We shall bet on cognitive extension instead of trying to re-create human intelligence --which is a legitimate area of research, but computer scientists should leave this to cognitive science and neuroscience.

    1. Last year, Page told a convention of scientists that Google is “really trying to build artificial intelligence and to do it on a large scale.”

      What if they're not? What if they're building an advertising machine to manipulate us into giving them all our money?

      From an investor perspective, the artificial answer certainly seems sexy while using some clever legerdemain to keep the public from seeing what's really going on behind the curtain?

    2. It seeks to develop “the perfect search engine,” which it defines as something that “understands exactly what you mean and gives you back exactly what you want.”

      What if we want more serendipity? What if we don't know what we really want? Where is this in their system?

  16. May 2021
    1. Turing was an exceptional mathematician with a peculiar and fascinating personality and yet he remains largely unknown. In fact, he might be considered the father of the von Neumann architecture computer and the pioneer of Artificial Intelligence. And all thanks to his machines; both those that Church called “Turing machines” and the a-, c-, o-, unorganized- and p-machines, which gave rise to evolutionary computations and genetic programming as well as connectionism and learning. This paper looks at all of these and at why he is such an often overlooked and misunderstood figure.
  17. Mar 2021
    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”

      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.

    1. Meanwhile, the algorithms that recommend this content still work to maximize engagement. This means every toxic post that escapes the content-moderation filters will continue to be pushed higher up the news feed and promoted to reach a larger audience.

      This and the prior note are also underpinned by the fact that only 10% of people are going to be responsible for the majority of posts, so if you can filter out the velocity that accrues to these people, you can effectively dampen down the crazy.

    2. In his New York Times profile, Schroepfer named these limitations of the company’s content-moderation strategy. “Every time Mr. Schroepfer and his more than 150 engineering specialists create A.I. solutions that flag and squelch noxious material, new and dubious posts that the A.I. systems have never seen before pop up—and are thus not caught,” wrote the Times. “It’s never going to go to zero,” Schroepfer told the publication.

      The one thing many of these types of noxious content WILL have in common are the people at the fringes who are regularly promoting it. Why not latch onto that as a means of filtering?

    3. But anything that reduced engagement, even for reasons such as not exacerbating someone’s depression, led to a lot of hemming and hawing among leadership. With their performance reviews and salaries tied to the successful completion of projects, employees quickly learned to drop those that received pushback and continue working on those dictated from the top down.

      If the company can't help regulate itself using some sort of moral compass, it's imperative that government or other outside regulators should.

    4. <small><cite class='h-cite via'> <span class='p-author h-card'>Joan Donovan, PhD</span> in "This is just some of the best back story I’ve ever read. Facebooks web of influence unravels when @_KarenHao pulls the wrong thread. Sike!! (Only the Boston folks will get that.)" / Twitter (<time class='dt-published'>03/14/2021 12:10:09</time>)</cite></small>

  18. Feb 2021
  19. Jan 2021
  20. Dec 2020
  21. Nov 2020
  22. Oct 2020
    1. Similarly, technology can help us control the climate, make AI safe, and improve privacy.

      regulation needs to surround the technology that will help with these things

    1. What if you could use AI to control the content in your feed? Dialing up or down whatever is most useful to you. If I’m on a budget, maybe I don’t want to see photos of friends on extravagant vacations. Or, if I’m trying to pay more attention to my health, encourage me with lots of salads and exercise photos. If I recently broke up with somebody, happy couple photos probably aren’t going to help in the healing process. Why can’t I have control over it all, without having to unfollow anyone. Or, opening endless accounts to separate feeds by topic. And if I want to risk seeing everything, or spend a week replacing my usual feed with images from a different culture, country, or belief system, couldn’t I do that, too? 

      Some great blue sky ideas here.

    1. Walter Pitts was pivotal in establishing the revolutionary notion of the brain as a computer, which was seminal in the development of computer design, cybernetics, artificial intelligence, and theoretical neuroscience. He was also a participant in a large number of key advances in 20th-century science.
  23. Sep 2020
  24. Aug 2020
  25. Jul 2020
  26. Jun 2020
    1. each of them flows through each of the two layers of the encoder

      each of them flows through each of the two layers of EACH encoder, right?

    1. It made it challenging for the models to deal with long sentences.

      This is similar to autoencoders struggling with producing high-resolution imagery because of the compression that happens in the latent space, right?

    1. it seems that word-level models work better than character-level models

      Interesting, if you think about it, both when we as humans read and write, we think in terms of words or even phrases, rather than characters. Unless we're unsure how to spell something, the characters are a secondary thought. I wonder if this is at all related to the fact that word-level models seem to work better than character-level models.

    2. As you can see above, sometimes the model tries to generate latex diagrams, but clearly it hasn’t really figured them out.

      I don't think anyone has figured latex diagrams (tikz) out :')

    3. Antichrist

      uhhh should we be worried

    1. We only forget when we’re going to input something in its place. We only input new values to the state when we forget something older.

      seems like a decision aiming for efficiency

    2. outputs a number between 000 and 111 for each number in the cell state Ct−1Ct−1C_{t-1}

      remember, each line represents a vector.

  27. May 2020
    1. Mei, X., Lee, H.-C., Diao, K., Huang, M., Lin, B., Liu, C., Xie, Z., Ma, Y., Robson, P. M., Chung, M., Bernheim, A., Mani, V., Calcagno, C., Li, K., Li, S., Shan, H., Lv, J., Zhao, T., Xia, J., … Yang, Y. (2020). Artificial intelligence for rapid identification of the coronavirus disease 2019 (COVID-19). MedRxiv, 2020.04.12.20062661. https://doi.org/10.1101/2020.04.12.20062661

  28. Apr 2020
    1. Abdulla, A., Wang, B., Qian, F., Kee, T., Blasiak, A., Ong, Y. H., Hooi, L., Parekh, F., Soriano, R., Olinger, G. G., Keppo, J., Hardesty, C. L., Chow, E. K., Ho, D., & Ding, X. (n.d.). Project IDentif.AI: Harnessing Artificial Intelligence to Rapidly Optimize Combination Therapy Development for Infectious Disease Intervention. Advanced Therapeutics, n/a(n/a), 2000034. https://doi.org/10.1002/adtp.202000034