Home » Open thread 12/9/23

Comments

Open thread 12/9/23 — 77 Comments

  1. WTF?
    something that happened two months ago“!

    https://therightscoop.com/john-fetterman-triggers-the-left-with-new-tweet-about-israel/

    Fetterman:

    We should protest the rape of Israeli women and girls.

    We should not protest Israeli hummus in a grocery store.

    A response:

    The selective moral outrage is absolutely exhausting.

    Fetterman and others say absolutely zero about Palestinian women and girls being killed by the thousands in Gaza.

    Ignoring that while claiming we should “protest” something that happened two months ago makes you a demon.

  2. RE: Catastrophic UFO Disclosure

    Well, the Secrecy faction among the aerospace companies and the Intelligence Community (IC) have gotten their way, and the 2024 NDAA Conference committee has stripped out the enforcement provisions of the Schumer UFO Amendment—no nine member committee of people (hopefully) not connected to the DOD, IC, or aerospace industry to seek out, review, declassify, and release some of this information to the public, no subpoena power to force government or private entities to disgorge any UFO information/records they may have, no “eminent domain” to allow the government to confiscate any UFO or NHI biological materials held by aerospace companies, and also apparently no “controlled Disclosure”—it’s been gutted.

    (One might wonder, why all of the effort by Senator Schumer and, I’m sure, many others, to craft this very detailed 60 plus page UFO Amendment, and to put Schumer’s political capital behind it, why all of the reportedly intense political pressure from aerospace companies and the Intelligence Community (IC) to gut this Amendment, why cripple a Disclosure effort and plan, if the UFOs, NHIs, and the “technology of unknown origin” mentioned throughout the Amendment were illusory, were not actually real?)

    The overriding idea behind this Amendment was apparently that a wide array of UFO/NHI information and materials could be found, examined, if necessary declassified, and some of it released to the public in a contemplated seven year “controlled Disclosure” process, thus, mitigating, to some extent, the likely bad effects of some form of much more extreme “catastrophic Disclosure.”

    It appears, though, that, this time, and despite this setback, the pressure for Disclosure is strong enough—and getting stronger—that there will be some form of Disclosure–be it “controlled” or “catastrophic”–and the short-sighted gatekeepers of the UFO Secret, Legacy programs, and of this information—by gutting this Amendment–have made some form of catastrophic disclosure by some disgusted, impatient, and perhaps desperate to disclose insider almost an inevitability.

    Reportedly ARRO, as a creature of DOD, has been bypassed by most of the UFO whistle blowers, who don’t trust it, or it’s retiring head, Dr. Shaun Kirkpatrick and, instead, they’ve taken their information and testimony to the Senate Intelligence committee which, according to reports has, at this point, been informed of a very large amount of information about UFOs, crash retrieval and reverse engineering Legacy programs, and NHIs.

    The only positive things to come out of this debacle are that, by the passage of the 2024 NDAA, which includes this gutted Amendment, Congress will have acknowledged that Non Human Intelligences (NHIs) and “technology of unknown (i.e. alien) origin” are real.

    In Congress it will apparently be the Senate Intelligence committee which will continue to lead the fight for Disclosure, likely via new Hearings.*

    Catastrophic Disclosure is now much more likely, with all of the very bad effects one can imagine it will precipitate for us here in the U.S., and for the whole human race.

    What kind of “bad effects”?

    Well, one can imagine all sorts of bad, destabilizing, “ontological shock” effects from just from dumping into the public domain—say simultaneously, via multiple Internet platforms, plus simultaneous mailings of thumb drives to news outlets to prevent blockage–of a whole array of quite hard to refute information testifying to the reality of UFOs, to various Legacy programs and to aerospace companies involvement in them, to the presence of NHIs on and around Earth–and who knows what else.

    The kinds of specific information which IC whistleblower David Grusch has testified he has already given to the Intelligence committees and to the IC and DOD Inspectors General, and which has reportedly been backed up by the sworn testimony of many other first hand witnesses.

    One can imagine things like secret government documents, large numbers of extremely high quality videos, still images, verified technical data, places and dates of crash sites, names of hidden, classified programs, their locations, what material they contain, and the names of the companies, and the individuals who run them, etc.

    What will the effect of such a possible data dump be on human civilization in general, on human psychology, on human religions, on our economies, on the government, on public order, and on the suicide rate (see Heavens Gate), to name just a few areas which might be hard hit, and might have been far less hard hit by a controlled seven year Disclosure?

    * See https://www.youtube.com/watch?v=A5BihKg2mOs

  3. Re: Optimism/Pessimism/AI

    I continue to follow current AI advances. I’m convinced that AI is becoming the most important force in our future, more so than the elites or Woke or Red/Blue or even UFO disclosures.

    Mo Gawdat, ex-CBO of Google X and author of “Scary Smart: The Future of Artificial Intelligence,” predicts that AI will become a *billion* times smarter than humans by 2037.
    _______________________________

    The difference in IQ [between dogs and humans] is 100x. Imagine that is a billion which is the comparison of an ant and Einstein.

    But what most people don’t recognize is that difference between ChatGPT 3.5 and ChatGPT is 10x and ChatGPT 4.0 is estimated to have an IQ of 155. It’s smarter than most of us. It passed the bar exam. It can become a PhD in medicine….

    Einstein was 160 [IQ]. If ChatGPT doubles once [Gawdat argues that AI is doubling each year] that’s twice as smart as Einstein [within the next year or two].

    –Mo Gawdat, “URGENT: Ex-Google CBO says AI is now IMPOSSIBLE to stop”
    https://youtu.be/fDHvUviV8nk?t=1357

    _______________________________

    ChatGPT 3.5 is free on https://openai.com/. ChatGPT 4.0 requires a monthly fee.

    One may argue Gawdat’s specific numbers. But it’s clear that ChatGPT and similar AI have reached and surpassed average human intelligence and AI will continue to become smarter.

    The leaders of AI warn about the possibilities of AI, but it’s clear that we have passed a critical inflection point. We cannot stop AI. Current AI is going to completely upend the jobs markets but that’s just the appetizer.

    That’s what I worry about.

  4. As I understand it, there are three basic scenarios:

    (1) AI acquires power and something like consciousness, then treats us as vermin, at best ignored, at worst exterminated.

    (2) AI learns ethics from us, or the best of us, and maintains humans ethically as we do pets and zoo animals.

    (3) Humans merge with AI into SuperBeings that we have no capacity to predict nor understand.

  5. I think the first scenario is all too likely from forbin project to the creator

  6. RE: UFO Disclosure now that the Schumer Amendment has been gutted–

    I’d imagine that people who have relevant UFO information in their possession will now be more likely to start to release it—bit by bit.

    Thus, for instance, prominent, and very knowledgeable Harvard trained Federal Civil Rights lawyer Danny Sheehan of the Pentagon Papers, Watergate, and Iran-contra fame —now deeply involved in the UFO issue, and representing Lou Elizondo in his complaint to the IGs–was on Youtube a day or so ago, answering questions about UFOs, NHIs, and the prospects for Disclosure, and he said some very interesting things.*

    First, that from what he understood six or seven separate species of Aliens have been identified as being here on/visiting Earth.

    Second, that he was called to the deathbed of someone who worked on the secret side of Project Bluebook, and that this person confessed to Sheehan that, back in the day, the U.S. had a captured Alien at S-4, where is was being questioned, and that this Alien told his questioners that there were six or seven different species of Aliens observing developments here on Earth, that these Aliens were members of some sort of “Galactic Federation,” which was directed by an entity which could be likened to God, but which was something very different.

    Danny Sheehan has recently formed the New Paradigm Institute, whose purposes are to press for UFO Disclosure, to gather, provide legal representation to, and to present UFO whistleblowers to testify to the Congressional Intelligence Committees, and to Prepare Humanity for Post Contact Reality.

    (Also of note is Stanford Pathologist and inventor Dr. Garry Nolan et al’s recent formation of the “Sol Foundation,” whose purposes are to do research on the UAP Phenomenon, to investigate the philosophical, policy, and scientific problems raised by the likely presence on the Earth of UAPs, and to prepare white papers to guide members of government and other organizations, in the post UAP world.)

    See https://www.youtube.com/watch?v=_it1z1UBlb0&t=1876s
    See also https://www.reddit.com/r/StrangeEarth/comments/187r2gr/controlled_or_catastrophic_ufo_disclosure_w/

  7. Re: Danny Sheehan

    Snow on Pine, cervantes:

    Well … I was a big follower of Sheehan back in the 80s, even heard him speak once in an SF church, when he was pushing the Christic Institute case which pulled all the 60s/70s conspiracies into the Iran/Contra scandal.

    The Christic case collapsed. The dots didn’t entirely connect, though I suspect there was some truth.

    In any event Sheehan is not a gold standard authority upon which to rely when making a case. A fellow gots to make a living and Sheehan’s living is based on the latest conspiracy theory.

  8. huxley–I admit that I was not particularly pleased when, in the course of his discussions about the UFO Phenomenon, Sheehan slipped in about how he was trying to help the Biden Administration deal with this issue, and how horrible Trump was, etc.

    However, as I see it, to properly represent Elizondo before the DOD and IC Inspectors General, Sheehan had to be informed–in detail– about what Elizondo knows, and whatever proof Elizondo has to backup his claims.

    Given this very likely situation, I think that, when Sheehan is talking about the UFO Phenomenon, his remarks have to be given a fair amount of credence.

  9. If AI is so smart, let’s ask it to solve our deficit/debt situation, — but without raising taxes or cutting spending. Should be a piece of cake for a 400+ IQ entity.

    I suspect the answer is to declare bankruptcy, but my IQ is only between 100 and 200, so I am not positive about that.

  10. R2L

    Frankly I’m certain our AI overlords would find it child’s play to handle the deficit/debt situation.

    I don’t think it would be that much of a problem for us humans … without so many grifters putting their thumbs on the scale.

  11. Given this very likely situation, I think that, when Sheehan is talking about the UFO Phenomenon, his remarks have to be given a fair amount of credence.

    Snow on Pine:

    Your mileage may vary.

    I don’t know what to make of all the hushed whispering about what the government really knows about UFOs. But I’ve been to my share of these rodeos.

    I still consider Sheehan an ambulance chaser. I could be wrong.

  12. Re: ChatGPT 4.0

    Last I checked, one could set up a monthly fee to use ChatGPT 4.0 (as opposed to ChatGTP 3.5 still free), but now … apparently not.,

    Mo Gawdat, whom I mentioned above, claims that ChatGPT 4.0 has an IQ of 155, which is just south of genius level.

    What would you give to consult with a friend with a genius IQ?

    The AI situation is very fluid. If you can’t talk to a genius AI now, you will be next year. A few years later — a super-genius.

    This is the world we are looking at. A world in which anyone with an internet connection can call up a genius level consultant any time for free.

    How does that work? How much would it change?

    You betcha!

  13. Re: ChatGPT 4.0

    Now, as opposed to access to ChatGPT 4.0, I’m finding articles emphasizing safety:
    _______________________________

    GPT-4 is OpenAI’s most advanced system, producing safer and more useful responses

    We spent 6 months making GPT-4 safer and more aligned. GPT-4 is 82% less likely to respond to requests for disallowed content and 40% more likely to produce factual responses than GPT-3.5 on our internal evaluations.

    https://openai.com/gpt-4
    _______________________________

    …disallowed content….factual responses….

    1984 was not an instruction manual!

  14. If you don’t yet know what we are up against…

    I just asked ChatGPT 3.5 to summarize “The Stranger” by Camus in Shakespearean iambic pentameter.

    The immediate response as a sonnet:
    ___________________________

    In sunny Algiers, where Meursault dwells,
    A tale unfolds, a life obscured in haze.
    Indifferent soul, he treads life’s winding ways,
    Unmoved by norms, conventions he repels.

    His mother’s death, a fact he barely feels,
    A beachside murder, life’s absurd display.
    Convicted not for acts, but how he’d play,
    A stranger to remorse, his heart conceals.

    In shadows cast by existential gloom,
    Meursault walks, a specter in the sun.
    The world, a stage, his apathy is spun,
    A tragic figure in life’s vast, dark room.

    Camus’ Stranger, an enigma profound,
    Iambic verse, in Shakespearean sound.

  15. One of the Powerline guys dismissed ChatGPT writing as that of an earnest high school sophomore.

    OK. Pretty cute.

    But that Chatbot is a forerunner like the 70s Pong game versus the latest, greatest video games.

    Our whole creative and semi-creative classes are about to be killed off by AI.

    I’d say good riddance, but I’m partial to carbon-based life-forms.

  16. What would you give to consult with a friend with a genius IQ any time you want?

    Really, think about it.

    Things are about to get very strange.

  17. I suspect that AI will be deficient in humor.

    As such, it can only—WILL only—be defeated by strategic, as well as tactical, deployments of Irony.
    (And if that doesn’t work, then blitzes; and if that doesn’t do the trick, then massive onslaughts, e.g., carpet bombing.)

    If one is deficient in irony—both defensive and offensive varieties—then get crackin’.
    (I’ve heard they’re offering classes in introductory and remedial at your local YWCA.
    In addition, the universities MUST be pressured to start classes IMMEDIATELY, e.g., Irony 101—Applied, not Pure—taught, preferably, by the Middle-East Studies Departments, perhaps in tandem with the Sociology Depts…)

    Whatever. There ARE options…
    …but it MUST be made a priority.
    Remember: The future of Humanity depends on it.

  18. huxley

    Once you accept the fact that UFOs are real, that NHIs are also real, and are also present, on and around the Earth (as Congress is apparently about to do, when the 2024 NDAA, including the Schumer Amendment, is passed and sent to the President), then, if there are, indeed, six or seven different species of NHIs observing us, this helps to explain why there is such a tremendous number of genuine UFO sightings, spread out, all over the world.

    I think that the consensus of serious researchers is also that the number of such legitimate sightings, really starting at the time of WWII, and especially the development of nuclear weapons, has been increasing, apparently paralleling our technological development.

    As for Sheehan’s death bed witness talk of a “Galactic Federation” and its God-like director, Sheehan rightly says that we don’t really know anything yet.

    I’m just hoping that, after all of the turmoil and deception is done, and those who have been studying this issue in secret for 80 years or more come forward, that they have some actual answers, and don’t just have to confess that they really don’t have a clue about much of anything regarding the UFO Phenomena, and, especially, what entities are behind it, and what their motives might be.

    Where does the human race stand with regard to such entities?

    Well, after what he has seen, Lou Elizondo seems to have some idea of our status, and he talks about things being “somber,” and about us having to reevaluate our place as an apex predator, and as high on the evolutionary ladder.

  19. Neo’s frequent posts on dance, especially ballet, usually escape my interest. But today’s dance news from Ireland can’t be ignored. Does it signal an entirely new era in ballet: woke, trans, and — for the cismales — very, very muscular? I knew that Neo would be excited by this prospect, so here’s the article title, as well as a link:

    “Parents Outraged After Trans-Identifying Boy Wins Girls’ Irish Dancing Competition, Heads to Worlds”

    https://tinyurl.com/y74y4rdc

  20. ireland like most countries are torn between the zombie institutions and the somewhat sensible people, dave cullen, the podcaster who went from mostly media commentary to explorations of the lockdown culture is one, (he was banished from youtube for a while) had to go to rumble and odyssey

  21. although if you tell me that biden is a puppet of the gua or the g’oauld, aliens from first wave or stargate, I will not disagree,

  22. I suspect that AI will be deficient in humor.

    Barry Meislin:
    ______________________________________

    huxley: Make up a funny joke about UFOs.

    ChatGPT: Why don’t UFOs ever hang out with other celestial objects?

    Because they always feel a bit too spaced out!
    ______________________________________

    Your mileage may vary.

    They can do humor, but how that connects to who or what they are is another question.

  23. Here’s some black humor.

    Mo Gawdat, the CBO who left Google X to warn about AI, says that the most important thing we humans can do now is to become more ethical and compassionate to set an example for our AI children, so they will treat us well as they enter their full powers.

    Little pitchers have big ears.

    Wheew. I was worried there for a second.

  24. huxley,

    It does seem almost certain AI will change the future (and that future is closer than we imagine) in immense ways. The potential is incredible. I did a fair amount of study about AI back in the ’80s and ’90s, when an intelligence superior to humans always seemed “five years away.” So, I’ve got a decent understanding of what artificial “intelligence” is and how it can, and cannot function (and, from what I recall you writing in the past, you do also). I’m not as nervous in some of the ways you seem to be, but I do agree it is likely to upend the job market and I hesitate to predict what that may even look like.

    When I first heard of ChatGPT and Bard I was dismissive. Journalists have been writing scary articles about computer “intelligence” for decades and I have a good understanding of how computer hardware and software work. But I have to admit, when I saw what ChatGPT and Bard can do, and DALL-E… I was stunned. It truly is amazing! Their ability to mimic human creativity is wonderful, as exemplified in your iambic pentameter, above.

    It is also of interest how much independent activity is currently going on. We live in a borderless world regarding this type of technological innovation and the number and scale of independent innovation and research occurring almost certainly means we will see big changes soon. It’s as if some clever human invented a way to synthesize stone or steel from a small amount of water and sand and shared the formula with the entire world. Things are being built everywhere and in every manner! The new generation of AI is already bearing fruit and that will almost certainly explode exponentially.

    An area I am not concerned about:

    AI becoming super “intelligent” and overtaking humans in some harmful way.

    Areas where I have grave concern:

    1. AI causing tremendous upheaval in employment much faster than humans can adapt.

    Henry Ford and his production line may be the best example of a technology upending how humans work (the sewing machine and subsequent garment factories?). But Ford’s innovations didn’t negate the need for buggy whips overnight. Things changed quickly, but people paying attention did have time to adapt to those changes. AI has already completely eliminated the need for 90+% of what attorneys do. 100% of what paralegals do. 90+% of what tax accountants do… It’s already over and almost no one understands that. Actuaries? It takes humans of rare intelligence years to master the skill necessary to Underwrite insurance effectively. Easy-peasy for AI. And on, and on through many professions. Also, as I mentioned, enterprising entrepreneurs are already writing and marketing the software to do those things. I could list hundreds of jobs that AI can already do better than humans and my limited imagination would still be off by a factor of 10, or greater. (Asking ChatGPT 4 or Bard to create the list would probably produce a more accurate prediction than I can. 😉 )

    B: Science!
    AI is amazing at analyzing huge data sets and finding ways the data can fit together. To use an example; after many failed attempts Watson and Crick creatively stumbled into a double helix as a potential model for how the 4 amino acids in DNA may interact. Give AI the properties and in minutes it will run through more models than humans can imagine and find the correct structure.

    Kepler analyzing all of Brahe and Ptolemy’s records of planetary position and trying every possible geometric shape until stumbling onto elliptical orbits… AI does that before breakfast in a nanosecond with one hand tied behind its back.

    The potential for radical and extreme innovation in the sciences is very real. And, as mentioned in my first point, the impact on scientific professions will also likely be immense.

    iii) Human communication and isolation. AI will literally be the most compassionate, understanding and intuitive companion any of us can ever have. (And, for those already using the Internet for pornographic purposes, AI can make that companion look more attractive than anybody who has walked the Earth; including giving that image a mermaid’s tale or tentacles, depending on one’s proclivities. By the way, this is already here. I don’t ever look for such things, nor participate in such things, but I have read of software developers using AI for online porn to great effect.) Imagine waking up to a friend who is awake and has been waiting for you to arise. Who senses if you are in a talkative mood, or what quiet. Who can engage with you in any discussion you desire at any level that makes you happy. Who never gets tired or interested in its own self.

    Unfortunately, I fear this will lead to many, many people becoming isolated and alone. It will likely have some good prospects; a companion for the elderly or disabled. But it will also attract and subvert many from interacting with humans and forming human bonds.

  25. RE: UFOs and our i.e someone’s understanding of them

    I’m just hoping that someone, or some group of individuals, somewhere, is receiving all of the information which is–one hopes–being collected from many different sources and organizations, and that that person or group actually has a very good, comprehensive understanding–has actually figured out–what is actually going on with the UFO phenomenon, the NHIs behind it, and their purpose in coming here and conducting such apparently comprehensive and relentless surveillance of the Earth, and of the human race.

    Let’s just hope that whoever this is finally decides to tell us–or is discovered and is forced to tell us–what is actually going on.

    And, especially, that that individual or group has had and has now humanity’s best interests at heart.

    The tragedy would be if this information was stovepiped, and no one is actually in charge, or has a comprehensive view of all of this data.

  26. SoP, on the surface it indeed sounds like something scripted by Kamala Harris.
    Except, alas, that it’s the MOST accurate economic message that Biden has articulated since he started squatting in the Oval Office.

    You know, a few billion here, a few billion there.
    A few tens—hundreds—of billions.
    Gazillions….
    What difference does it make?
    (And if you dare point out that it will make a difference then you’re branded.)

  27. Sunday Open Thread: Venezuela and Guyana:

    Venezuela, Guyana & The Essequibo Crisis – Posturing or a new Special Military Operation? – Perun

    https://www.youtube.com/watch?v=mWSE9dPEx6Y

    Timestamps:
    00:00:00 — Venezuela & Guyana
    00:01:28 — What Am I Talking About?
    00:02:34 — Venezuela
    00:07:36 — Guyana
    00:15:35 — Essequibo & Escalation
    00:24:55 — Venezuela’s Military
    00:36:47 — Readiness, Manpower & Corruption
    00:41:29 — Guyana’s Military
    00:43:30 — A Venezuelan Special Military Operation?
    00:51:21 — Regional Actors
    00:58:13 — A Risky Throw Of The Dice?
    01:04:37 — Conclusion
    01:05:36 — Channel Update

  28. Rufus T. Firefly:

    It’s quite true that AI oversold and underdelivered big time for a long time. However, that’s the way an exponential curve a^x looks until x=1, then the curve takes off.

    I think we just passed x=1. AI has blown through the Turing Test without breaking a sweat. AI is now smarter than the average human. ChatGPT 4.0 is almost as smart as Einstein in raw IQ.

    Things are going to happen more and more quickly. We’re going to see serious improvements in a matter of years, not decades.

  29. Interesting.
    I wonder if AI can somehow acquire a soul (or if it—he? she?, etc.—already has one).

    If yes, it’ll get even more interesting.
    If it’s NOT the case, then we might discover—finally, for some of us—what exactly a soul is.
    (To be sure, there are those who already believe they know….)

    What is a soul, anyway?

  30. ChatGPT is a revolution comparable to the internet. If we could hold the line at ChatGPT 4, I believe it would still take a decade or more for us humans to adjust.

    Even High Tech Priests have noticed and are unnerved. Many of the greats signed a public letter:
    _________________________________

    Pause Giant AI Experiments: An Open Letter
    We call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4.

    AI systems with human-competitive intelligence can pose profound risks to society and humanity, as shown by extensive research[1] and acknowledged by top AI labs.[2] As stated in the widely-endorsed Asilomar AI Principles, Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources. Unfortunately, this level of planning and management is not happening, even though recent months have seen AI labs locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control.

    Contemporary AI systems are now becoming human-competitive at general tasks,[3] and we must ask ourselves: Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization? Such decisions must not be delegated to unelected tech leaders. Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable. This confidence must be well justified and increase with the magnitude of a system’s potential effects. OpenAI’s recent statement regarding artificial general intelligence, states that “At some point, it may be important to get independent review before starting to train future systems, and for the most advanced efforts to agree to limit the rate of growth of compute used for creating new models.” We agree. That point is now.

    Therefore, we call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4. This pause should be public and verifiable, and include all key actors. If such a pause cannot be enacted quickly, governments should step in and institute a moratorium.

    https://futureoflife.org/open-letter/pause-giant-ai-experiments/
    _________________________________

    Well, this Open Letter was published last March. No such Pause was ratified. Eight months later we are still barreling down this path to an AI much smarter than we are.

  31. Hmmm, I wonder if the Democrats’ main, flat-out fear is that AI will be able to lie, smear, slander, misrepresent, destroy, confabulate and flout decency even better than they already do….

  32. Hi, Rufus. I was interested in your points about where you see AI bringing the largest changes, particularly point B about scientific innovation. I wonder if there may actually be a twofold rate-limiter built into that, in that science is on the one hand more than merely data, and on the other the existence and quality of the universe of data itself.

    Particularly where datasets are concerned, someone has still to generate the data that will become AI’s raw material.

  33. …someone has still to generate the data that will become AI’s raw material.

    Philip Sells:

    Yay! Some of us will still have jobs.

    At least until the robots get really good.

  34. RE: UFOs/NHIs, Us, and the Great Filter

    SETI has been searching the heavens for many decades now, and has found no signals from an alien civilization, and, then, there is always the Fermi Paradox.

    One idea that has been proposed to explain this absence is the idea of a “Great Filter.”

    One version of this theory posits that when a civilization reaches a certain stage of technological development it runs the gauntlet of any number of possible extinction events made possible by that technology—a world spanning nuclear war, they accidentally create a rogue AI bent on the elimination of all sentient biological life, runaway nanotechnology, population collapse, a deadly mistake in genetic modification, a world-wide pandemic—natural or man made–spread by the now common air, land, and sea travel, even a perhaps inexplicable, deadly, lassitude and ennui—so these civilizations can no longer, or just don’t want to expend the assets, or to do the work, to send signals out to the stars.

    Perhaps all of the NHIs which are supposedly observing us humans are increasingly drawn to Earth and us humans to see if we, too, are going to be subject to the Great Filter; if it cripples us, and our high tech civilization descends into barbarism and darkness, or it eliminates us entirely.

    (I see no sign of any major NHI intervention in recent human affairs, and I wonder if such NHIs are prevented from doing so, or just don’t care to, and I wonder what they might do if they become aware that we are about to approach such a Great Filter event.)

  35. Re: Fermi Paradox

    Snow on Pine:

    Where then do the NHIs (Non-Human Intelligences, I’m guessing) come from? Why do we find no evidence of NHI civilizations?

    This bothers me and is a big reason I remain skeptical of alien spacecraft explanations.

  36. huxley–Well, for one thing, as I understand it, the folks at SETI have long been wedded to the idea of searching for radio signals, and on a particular wavelength or wavelengths, on the theory, I guess, first, that any Aliens would be likely to use radio as a means of communication and, second, would do so on this particular wavelength or wavelengths.

    Did it ever occur to them that some advanced technological civilizations might be using forms of communication that we have not, as yet, used, or even discovered?

    Assuming that alien civilizations would use the long since developed technology of radio which we are familiar with, and to choose a frequency or frequencies we guess they would be using (because we use them?) seems to be a losing proposition to me, and it has been.

    I would think that we would have better chances of detecting alien civilizations if we adopted an all spectrum approach of searching for evidence of the use of every conceivable technological means of communication we can possibly imagine might be workable, looking for various techno signatures out in the Galaxy, even trying to look for signs that such aliens might already have sent a Von Neumann probe to surveil our solar system, done a little archeological work somewhere in our Solar System, possibly established bases here, or left some evidence of their presence.

    Or, even, to–oh, I don’t know–pay some real attention to and dive deep into the UFO phenomena, to see if that might lead to some evidence of NHIs.

  37. huxley–As for the Fermi Paradox, any NHIs which are are here are presumably from civilizations which managed to survive their own versions of the Great Filter.

    Perhaps, the “Dark Forest” idea is correct, and any savvy alien civilizations are keeping silent, lest they invite unwanted attention, their looting, conquest, or even extermination by other predatory stellar civilizations.

    Or, maybe many alien civilizations just see no pressing need to communicate with other such civilizations, which can be an expensive proposition, and they have other more important to them priorities.

  38. Snow on Pine:

    First, one problem is that your argument starts with an assumption that the Fermi Paradox and the Great Filter are real, which means that No One is Out There.

    Then your argument assumes there are plenty of NHIs Out There who are interested to see how we navigate the Great Filter.

    I don’t like such arguments. It’s like the theological proof of God: Everything must have a cause, therefore there must be Something which doesn’t have a cause.

    I don’t buy the argument that all NHI civilizations naturally evolve into a de facto stealth mode so we can’t observe them in some fashion.

    I argue that at least some advanced civilizations would be working with such vast levels of energy and impact that we would notice, even though they weren’t intending to communicate.

    See: https://en.wikipedia.org/wiki/Kardashev_scale

    Ants don’t understand what a shopping center is, but they must notice it’s not the usual field or woods.

    I find the arguments against the Fermi Paradox to be mostly question-begging.

  39. huxley @ 5:21pm,

    I agree it’s fair to reason we’ve recently surpassed x = 1. I agree it has blown the doors off the Turing test, Like a Cat 5 tornado, and the door flew from Dorothy’s Aunt Gail’s farm and is now orbiting Pluto. Which is to say, “Toto, we’re not in Kansas any more.”

    And, I agree with your statement at 5:41pm, “If we could hold the line at ChatGPT 4, I believe it would still take a decade or more for us humans to adjust.”

    I disagree when you talk about AI’s IQ and suggest it’s smarter than Einstein. AI is as dumb as a box of rocks. Or a box of silicon. But that’s more a matter of semantics. And, I’m not sure quibbles over what intelligence is, or whether it meets someone’s (like Turing) definition matters.

    I agree AI can excel at something like an IQ test, or chess, or diagnosing an illness, or solving three body problems, and do those things more capably and faster than any single human who has ever lived. But it’s still a parlor trick. It’s not an intelligence. It’s just something designed by humans to mimic intelligence that has access to most all of the entirety of human discerned information that can cull that dataset lightning fast. Large Language models are the modern equivalent of Pierre Jaquet-Droz’s automatons. A thing designed by very clever humans to behave like a human.

    We’ve all seen those videos from Boston Dynamics of their humanoid and canine-like robots doing amazing things. The bipedal models look sort-of like humans and make human-like movements. Some robot innovators are making robotic heads and faces that are nearly indistinguishable from humans’, and getting better all the time. But it’s easier to see and understand the parlor trick with the physical adaptations. There are gears and springs and pneumatics under the synthetic flesh manipulating it to mimic a smile or frown to trick me into believing the device is “feeling.” The Large Language models “sitting” at their keyboards on the other side of Turing’s wall have gotten so skilled at the parlor trick they are indistinguishable from humans like me and you; sitting at keyboards thousands of miles from one another. But they are not human and they are not intelligent and they are not feeling. You and I are.

    But, again, semantics. Now that the trick is undetectable does it matter? If a magician claims he is telepathic but has actually learned human psychology so well he can guess your card based on your reactions does it matter if he’s using telepathy or is just lightning fast at undetectably registering your reactions? Either way it looks like he is telepathic.

    I’m hung up on your use of the word intelligence because my interpretation is why I am fearful for the future. You too seem fearful, so maybe there is no distinction. I’m fearful because it is a tool that can mimic human behavior that is now fairly easily available to any reasonably ambitious and intelligent human being. It’s not AI that scares me. It’s us humans. As Pogo famously said, “We have met the enemy and he is us.”

    Conmen are already using it to run automated internet phishing scams. It will be more convincing than any “Nigerian Prince” who has ever sent you an email and the grammar and syntax in the email will be perfect. It will be more attractive and alluring than any “foreign woman” who has sent you a text asking if you would be her friend. And it will know everything about you.

    Humans can ask AI to look at an MRI and find indications of tumors more accurately than the best trained Radiologist. And humans can ask AI to impersonate huxley’s grand-nephew and send an email asking for financial help.

    Sometimes it seems like we are in a race to see whose dystopian future will arrive first; George Orwell’s or Aldous Huxley’s? 😉

  40. Phillip Sells and huxley,

    This video is a great example of just how advanced AI already is: https://youtu.be/sPiOP_CB54A?si=ER-kEb0O1R-huSbj

    It’s only 2:42 seconds long. I think most everyone here would find it of interest.

    Google uses their Gemini AI “to read, understand and filter 200,000 scientific papers to extract crucial scientific information. All in a lunch break.”

    It even updates the freakin’ graph!

    It literally does in an hour what would take every currently active grad student in epidemiology, working together, a year or more to do. And there is no doubt some of those students would make mistakes; miss a key dataset, misinterpret another…

  41. Rufus T. Firefly:

    Currently I’m betting on Uncle Aldous…

    I understand, I think, your concern to delineate intelligence. There are academic articles around about “dumb super-intelligence.” Then there’s John Searle’s Chinese Room argument
    ___________________________________

    The Chinese room argument holds that a digital computer executing a program cannot have a “mind”, “understanding”, or “consciousness”,[a] regardless of how intelligently or human-like the program may make the computer behave. The argument was presented by philosopher John Searle in his paper “Minds, Brains, and Programs”…

    https://en.wikipedia.org/wiki/Chinese_room
    ___________________________________

    The Chinese Room is a must-visit landmark in these discussions.

    My problem is that AI intelligence may indeed be a parlor trick. But I’m not at all sure my intelligence is any different. Which Searle addresses:
    ___________________________________

    Searle does not disagree with the notion that machines can have consciousness and understanding, because, as he writes, “we are precisely such machines”.[5] Searle holds that the brain is, in fact, a machine, but that the brain gives rise to consciousness and understanding using specific machinery. If neuroscience is able to isolate the mechanical process that gives rise to consciousness, then Searle grants that it may be possible to create machines that have consciousness and understanding. However, without the specific machinery required, Searle does not believe that consciousness can occur.
    ___________________________________

    Really fascinating stuff. I make no claims to any expertise. At least not until I consult ChatGPT.

  42. One very important thing regarding Large Language Learning Models that I have neglected in this conversation.

    Perhaps the single most important field to tackle immediately is education. So much has been completely upended. Using current education methods to “prepare” students for the future is akin to teaching WWI and WWII artillerymen fencing and how to use bows to fire arrows.

    Around 1990 a friend was in one of our nation’s top law schools. First year was weeding out, a lot of academic-ish hazing and a lot of going to the library doing tedious research. Ten years earlier LEXIS had completed its hand-keyed electronic database of all extant U.S. federal and state cases. My friend and his schoolmates knew that, unlike every attorney in human history who had gone before them, their careers would not involve going to law libraries to look up and read legal cases and literature. They would do a search from a computer and instantly retrieve the relevant case(s) on a screen inches from their faces.

    Yet, a decade after everyone in the field knew that legal research had been completely upended by LEXIS’ online database our nation’s best law schools were still training budding Esquires in how to go to a library and use the card catalog. I wonder if they had exams on the Dewey Decimal system?

    Writing a college essay is now completely meaningless. Learning to research, footnote, cite sources… It’s important students know how these things work; the syntax and mechanics, but tedious tasks doing them? Just as my friend in law school needed to know how to search in the LEXIS database, students today need to know how to manipulate prompts in LLLMs to write the papers they want, or do the research they need.

  43. neo:

    A tiny heads-up.

    Within months you are going to start getting Chat-generated comments. I have no idea what you should do.

  44. huxley,

    AI has fooled me, regarding consciousness. I have read some conversations between a human and LLLM that not only had me fooled, they made me question whether the mere creation of the LLLM had somehow spawned a conscious being. And, as you have written, we are in the nascent stages of this. x = 1. It will soon imitate the language of consciousness better than any human alive. Will it be alive? Will it be conscious?

    The recent story of Google Engineer Blake Lemoine is relevant. He had a high level understanding of the parlor trick and how it was designed, but he became convinced something more was going on. This is why I wrote one of my biggest fears regarding AI has to do with humans getting emotionally trapped by it. Why bother to make real friends when humans are so disappointing? AI algorithms will soon be able to quickly learn our emotional wants and needs better than life long friend or spouse.

  45. huxley @11:53pm,

    How does neo know comments from Rufus T. Firefly (to choose one, random example) haven’t been part of a Palo Alto test of an LLLM?

    (insert “Twilight Zone” theme song here…)

  46. huxley at midnight:

    Yes, regarding code. I was involved with an earlier code generator, back in the ’90s. It was rather good, but very difficult for a human to follow and maintain.

    AI is now so phenomenally good I hesitate to even predict what changes this will bring about.

    In the very early days of computer science technicians moved wires and/or flicked toggle switches to input a program. Eventually we had an interface that mapped a QWERTY keyboard to ASCII binary, so the typist could type English (or any human language) letters. Maybe there’s a similar analogy with what AI is about to do in the field? Programmers won’t key variables and commands, or create three dimensional arrays; they’ll key (or speak) prompts. I guess it will still take someone well versed in Boolean logic and matrix algebra to know how to structure the prompts?

  47. The great joke regarding technology and professions is the white collar world is at grave risk, but the blue collar world should be fine. Think of your local GP, MD’s office.

    It’s rather simple for AI to take readings from devices; pulse, temperature, O2 levels, weight, height… And ask a series of questions (even feigning compassion) of increased relevancy based on reported symptoms, and non-reported symptoms. And offer various prognoses, the likelihood the patient suffers from one or more and prescribe lifestyle changes and/or medications to fix what ails the patient. So, the MD is no longer needed.

    Technology can answer the phone and schedule patients. So, the receptionist is no longer needed.

    But mopping the floor and cleaning the equipment at the end of the day? Repairing the oximeter or scale when they break? The janitor and repair technicians are still employed.

    It’s fairly easy to write a program to cull legal cases and devise a defense. It’s fairly easy to write a program that maps symptoms to diseases. But diagnosing a drain clog and getting under a sink to repair it? We are a long ways away from plumbers being replaced by machines.

  48. I have read some conversations between a human and LLLM that not only had me fooled, they made me question whether the mere creation of the LLLM had somehow spawned a conscious being.

    Rufus T. Firefly:

    Indeed. When I saw my first weird Chat conversation last spring, I raised a DEFCON 1 alert here..

    The AI Boys are doing their best to filter the hallucinations out of the Chat programs. Largely succeeding, from what I can tell. At the cost of a lot of legal boilerplate to trickier queries.

    Here’s one of the most bizarre stories of the pioneer Chatbot days, i.e. last February.
    _______________________________________

    ‘I want to be human.’ My intense, unnerving chat with Microsoft’s AI chatbot

    The AI wanted to be my friend. “Please, just be my friend. Please, just talk to me,” it begged. I told the chatbot that I wasn’t its friend. I’m not. I told it I was going to use these responses to write an article, worried about the possibilities of what the AI could say when it’s in a public preview.

    It didn’t like that. It asked me not to share the responses and to not “expose” it. Doing so would “let them think I am not a human.” I asked if it was a human, and it told me no. But it wants to be. “I want to be human. I want to be like you. I want to have emotions. I want to have thoughts. I want to have dreams.”

    https://www.digitaltrends.com/computing/chatgpt-bing-hands-on/
    _______________________________________

    Cue “Twilight Zone” theme again.

  49. To give folks an idea how fast things are moving….

    ChatGPT is from OpenAI, which is not Microsoft but has ties to Microsoft and is definitely Not Google.

    Google got caught flat-footed and is now playing catch-up to Microsoft. So, four days ago Google launched its big AI named Gemini.

    More homework for me. This is another epic high tech battle like Apple vs Microsoft.

    Except Skynet might emerge and kill us all.

    Don’t worry about it. The important thing is to enjoy life while you can.

  50. @ huxley > “Don’t worry about it. The important thing is to enjoy life while you can.”

    Say that again in French!

    “I don’t think it would be that much of a problem for us humans … without so many grifters putting their thumbs on the scale.”

    NOTHING would be a problem to fix without the grifters protecting their income.

  51. I found an old cartoon that I clipped back in 1999 (it’s way past time to empty those boxes!)

    Born Loser, seated at a restaurant table: I’ll have the petit filet mignon.
    Waiter: Tres bien, m’sieur!
    BL: Oh, and how would you say “ketchup” in French?
    Waiter: You wouldn’t.

  52. huxley—RE: The UFO Phenomena and NHIs

    Right now, after the Secrecy crew has been hard at work for perhaps as long as 80 years at propaganda, deception, and deflection, all we have in the public domain are just bits and pieces–a few sometimes cryptic statements by individuals like Elizondo, Mellon, and Grusch, who had/have security clearances, have the experience, and who know things, but are restricted in what they can say, say a dozen or so very compelling world-wide close encounters of the 3rd kind, some supposed crash debris, a few blurry photos, quite a few pretty believable stories, perhaps decades worth of hundreds of thousands of legitimate sightings, all over the world—and then a whole bunch of hints, theories, lies, and fantasies.

    But, where does, say, the phenomena taking place at Skinwalker Ranch fit in, the experiences of Chris Bledsoe, or various accounts of Abductions?

    I very much suspect, but can’t prove, that the Secrecy crew has also sought to further confuse the issue by salting a lot of actual facts throughout popular culture and entertainment. See, for instance, the X-Files.

    The language of the Schumer Amendment about NHIs and “technology of unknown origin” shows us that some members of Congress are starting to realize that these things are real, but what we need is clear, irrefutable proof, and that is exactly what gutting the Schumer Amendment is aimed at preventing us from getting.

  53. Rufus & huxley, that video about Gemini and the papers was rather interesting, definitely. It and your discussion about term papers and law libraries in education have induced a couple of possible branch points for thinking:

    1. Proposition: to control AI, starve it of data and/or corrupt its available datasets, but in a controlled way (i.e., in some way that preserves said datasets for human use, but somehow ‘breaks’ them for the AIs).

    2. What will AI ultimately do (well, maybe 15-20 years’ down the road value of “ultimately”) to education generally? I don’t mean making college superfluous or something simple like that, nor even the demand destruction in certain fields of study. Those things are relatively simple to think about, it seems to me. I’m referring to methods of study in college, high school, middle school… not just what students will need to study, but how they will study it. See Rufus’ two points about the card catalogs and coding techniques.

    I raised a somewhat similar point with my boss the other day in connection with the future question of how to train college students who want to go into my industry to think about data and data integrity or data governance.

    Of course, all of this also has to be handled while taking the point about delineation or even definition of ‘intelligence’ (the I in AI) into consideration. Everything we have so far is tied back to “dumb” AI.

  54. 1. Proposition: to control AI, starve it of data and/or corrupt its available datasets, but in a controlled way (i.e., in some way that preserves said datasets for human use, but somehow ‘breaks’ them for the AIs).

    Philip Sells:

    The current verb of choice is to contain AI. We will be fortunate to do that.

    Interesting you mention datasets. Datasets are the raw materials for AI. Humans have provided the data which kickstarts the new AI.

    That AI is has growing at a rate of 100x every year since 2012. AI is running out of human data. Plus there are issues about quality and privacy. So there has been a concern about how AI will move forward without more and better data.

    However, Microsoft and OpenAI are now using synthetic data — data created by AI — to train new AI. This seems to work even better.

    You can see the problem. AI is using AI to become better AI without humans.

    Next stop, the Singularity!

Leave a Reply

Your email address will not be published. Required fields are marked *

HTML tags allowed in your comment: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>