SD Marlow

• SD MarlowNov 11, 2022, 5:39 AM

  1 point

  0

  on: Up­dates and Clarifications

  With the collapse of FTX Future Fund, I’m guessing this prize/​contest is no longer funded.

• SD MarlowNov 5, 2022, 5:49 PM

  0 points

  0

  in reply to: ChristianKl’s comment on: Re­view of the Challenge

  Errors from transposing or misreading numbers, placing the decimal in the wrong location, etc. The machine mind has a “perfect cache” to hold numbers, concepts, and steps involved. Math is just the simple example of their ability. Such machine minds will be able to hold every state and federal law in their mind, and could re-draft legislation that is “clean” while porting case law from the old to the new legal references.

  *for an example of current tech getting simple math wrong: https://​​twitter.com/​​KordingLab/​​status/​​1588625510804119553?s=20&t=lIJcvTaFTLK8ZlgEfT-NfA

• SD MarlowNov 5, 2022, 2:19 PM

  1 point

  0

  on: Re­view of the Challenge

  *You can’t align something that doesn’t really work (it not really working is the current danger). A better question is, can you take a working AI and brainwash it? The unhackable (machine) mind? A good system, in the wrong hands, and all that.

  **Again, free yourself from the monolithic view of current architectures.

Re­view of the Challenge

• SD MarlowNov 1, 2022, 5:32 PM

  1 point

  0

  on: “Cars and Elephants”: a hand­wavy ar­gu­ment/​anal­ogy against mechanis­tic interpretability

  I don’t see how mechanistic and modular are different, and it’s only clear (upon reading your last point) that your analogy might be flipped. If a car breaks-down, you need to troubleshoot, which means checking that different parts are working correctly. Interpretability (which I think fails as a term because it’s very specific to math/​xNN’s, and not high-level behaviors) happens at that level. Specific to the argument, though, is that these are not two different concepts. Going back to the car analogy, a spark plug and fuel injector are roughly the same size, and we can assume represent an equivalent number of neurons to function. ML fails by being focused on the per-neuron activity; and thus, AGI will be based on the larger structures. We would also ask what is wrong, in terms or elephant psychology, in trying to troubleshoot an elephants mind (how something “lights up” in a region of the brain from a scan is only a hint as to the mechanisms we need to focus on, though maybe the problem is that another area didn’t light up, etc).

  “Solving” AGI is about knowing enough about the modules of the brain, and then building from scratch. ML research is about sub-module activity, while safety seems to be above module activity. None of these things are compatible with each other.

• SD MarlowOct 26, 2022, 3:44 AM

  0 points

  −1

  on: Why some peo­ple be­lieve in AGI, but I don’t.

  First, we can all ignore LeCun, because despite his born-again claims, the guy wants to solve all the DL problems he has been pushing (and winning awards for) with more DL (absolutely not neuro-symbolic, despite borrowing related terms).

  Second, I made the case that amazing progress in ML is only good for more ML, and that AGI will come from a different direction. Wanting to know that seems to have been the claim of this contest, but the distribution of posts indicates strong confirmation bias toward more, faster, sooner, danger!

  Third, I think most people understand your position, but you have to understand the audience. Even if there is no AGI by 2029, on a long enough time scale, we don’t just reach AGI, there is a likely outcome that intelligent machines exist for 10′s of thousands of years longer than the human race (and they will be the ones to make first contact with another race of intelligent machines from halfway across the galaxy; and yes, it’s interesting to consider future AI’s contemplating the Fermi Paradox long after we have died-off).

• SD MarlowOct 17, 2022, 12:41 AM

  2 points

  0

  in reply to: exkn’s comment on: Why I think strong gen­eral AI is com­ing soon

  Advances in ML over the next few years as being no different than advances (over the next few years) of any other technology VS the hard leap into something that is right out of science fiction. There is a gap, and a very large one at that. What I have posted for this “prize” (and personally as a regular course of action in calling out the ability gap) is about looking for milestones of development of that sci-fi stuff, while giving less weight to flashy demo’s that don’t reflect core methods (only incremental advancement of existing methods).

  *under current group think, risk from ML is going to happen faster than can be planned for, while AGI risk sneaks-up on you because you were looking in the wrong direction. At least, mitigation policies for AGI risk will target ML methods, and won’t even apply to AGI fundamentals.

• SD MarlowOct 15, 2022, 3:06 AM

  1 point

  0

  on: “AGI soon, but Nar­row works Bet­ter”

  True enough that n AGI won’t have the same “emotional loop” as humans, and that could be grounds for risk, of some kind. Not clear if such “feelings” at that level are actually needed, and no one seems to have concerns about loss of such an ability from mind uploading (so perhaps it’s just a bias against machines?).

  Also true that current levels of compute are enough for an AGI, and you at least hint at a change in architecture.

  However, for the rest of the post, your descriptions are strictly talking about machine learning. It’s my continued contention that we don’t reach AGI under current paradigms, making such arguments about AGI risk moot.

• SD MarlowOct 13, 2022, 3:57 AM

  1 point

  0

  in reply to: Soren’s comment on: AI Timelines via Cu­mu­la­tive Op­ti­miza­tion Power: Less Long, More Short

  Computer vision is just scanning for high probability matches between an area of the image and a set of tokenized segments that have an assigned label. No conceptual understanding of objects or actions in an image. No internal representation, and no expectations for what should “be there” a moment later. And no form of attention to drive focus (area of interest).

  Canned performances and human control just off camera give the false impression of animal behaviors in what we see today, but there has been little progress since the mid-1980′s into behavior-driven research. *learning to play a video game with only 20 hours of real-time play would be a better measure than trying to understand (and match) animal minds (though good research in the direction of human-level will absolutely include that).

• SD MarlowOct 12, 2022, 3:29 PM

  1 point

  0

  on: My ar­gu­ment against AGI

  This is the boiled-down version of your argument? You clearly state that GOFAI failed and that classic symbolic AI won’t help. You also seem to be in favor of DL methods, but in a confusing way, say they don’t go far enough (and then refer to others that have said as much). But I don’t understand the leap to AGI, where you only seem to say we can’t get there, not really connecting the two ideas. Neurosymbolic AI is not enough?

  I think you’re saying DL captures how the brain works without being a model for cognitive functions, but you don’t expand on the link between human cognition and AGI. And while I’ve thrown some shade on the group dynamics here (moderators can surface the posts they like while letting others sink, aka discovery issues), you have to understand that this ‘AGI is near’ mentality is based on accelerating progress in ML/​DL (and thus, the safety issues are based on the many problems with ML/​DL systems). At best, we can shift attention onto more realistic milestones (thou they will call that goalpost moving). If your conclusion is that this AGI stuff is unlikely to ever pan-out, then yes, all of your arguments will be rejected.

• SD MarlowOct 11, 2022, 5:59 AM

  1 point

  0

  on: Ob­jects in Mir­ror are Closer Than They Appear

  Saying “a substantial amount of expertise is required” goes past what general actually implies (I use the example of 8th grade level knowledge, at which point, the system can branch in thousands of directions on it’s own).

  Saying that a good test of an AGI is for it to build a better AGI, while also saying that such an effort represents an immense risk?

Up­dates and Clarifications

• SD MarlowOct 11, 2022, 4:25 AM

  2 points

  0

  on: Don’t ex­pect AGI any­time soon

  Python syntax is closer to natural language, which plays into what the LLMs do best. I don’t think the “symbolic” aspect plays into this in any way, and that kind of misses the argument on symbolic reasoning (that the LLM’s are still just just doing correlation, and have no “grounding” of what the symbols mean, nor does any processing happen at that grounding level).

  I’m still confused by your position. Saying DL is capturing symbolic values (in this one case), but also that DL is going to fail (because...?).

• SD MarlowOct 9, 2022, 1:44 AM

  0 points

  0

  on: The Le­bowski The­o­rem — Char­i­ta­ble Reads of Anti-AGI-X-Risk Ar­gu­ments, Part 2

  Anyone talking about reward functions is not talking about AGI. This is the disconnect between brute force results and an actual cognitive architecture. DL + Time is the reward function for the posts doing well, BTW, because it fits the existing model.

• SD MarlowOct 8, 2022, 1:27 AM

  1 point

  −1

  on: The prob­a­bil­ity that Ar­tifi­cial Gen­eral In­tel­li­gence will be de­vel­oped by 2043 is Zero

  Heavy down votes in less than a day, most likely from the title alone. I can’t bring myself to vote up or down because I’m not clear on your argument. Most of what you are saying supports AGI as sooner rather than later (that DL just needs that little bit extra, which should have won this crowd over). I don’t see any arguments to support the main premise.

  *I stated that the AGI clock doesn’t even start as long as ML/​DL remains the de facto method, but that isn’t what they want to hear either.

• SD MarlowOct 4, 2022, 4:54 AM

  1 point

  0

  in reply to: Joaquin Cabezas’s comment on: Why I think strong gen­eral AI is com­ing soon

  Interesting, and not far from my take, which is that ML has been wearing AI as a skin (because built-in marketing). Now that it is “advancing,” it has to wear AGI as a skin to indicate progress. That AGI was originally an effort to step away from DL’s path, and return to something closer to original intent of AI as a field gets lost.

• SD MarlowOct 4, 2022, 4:39 AM

  0 points

  −4

  on: Why I think strong gen­eral AI is com­ing soon

  I guess I’m one of those #2′s from the fringe, and contributed my 2 cents on Metacalus (the issue of looking for the right kind of milestones is of course related to my post in relation to current challenge). However, I completely reject ML/​DL as a path toward AGI, and don’t look at anything that has happened in the past few years as being AI research (and have said that AI officially died in 2012). People in the field are not trying to solve cognitive issues, and have rejected the idea of formal definitions of intelligence (or stated that consciousness and being sentient or not isn’t an issue). I have to use AGI as a label for “stealth mode” reasons, but would prefer to separate field of study from the implementation side. And while I’m not trying to build better models of the human mind, I have come to understand consciousness in a fundamental way (and working from an embodied framework, it’s easier to see just how incapable current efforts are).

Frontline of AGI Align­ment

• SD MarlowMar 25, 2022, 1:14 AM

  1 point

  in reply to: Aron’s comment on: Visi­ble Thoughts Pro­ject and Bounty Announcement

  I was “archiving” the link to this page and thought I’d see what’s been going on. Updates seem to only be on the discord. Anyway, since they allowed me to post longer thoughts there, figured it would be fine for me to drop it here as well. https://​​sd-marlow.medium.com/​​slaying-the-ml-dragon-7ce0a2e4e3a6

  From your post, you’re looking at this in much the same way I was when I attempted to do a short run (to work the bugs out and really understand whats involved). However, “actual thoughts of the DM” is the wrong explanation for what they want. The examples of of what they are accepting look to be nothing more than the “common sense” stuff current ML models fail to capture (thus, explicitly stated in the runs). Also, from comments in the discord, it seems like the info captured is post-process, despite the desire for pre-prompt thoughts. Not trying to discourage; just showing my thinking on the process, and that it wasn’t what they wanted.

• SD MarlowDec 9, 2021, 8:10 PM

  1 point

  in reply to: SD Marlow’s comment on: Visi­ble Thoughts Pro­ject and Bounty Announcement

  Using scenes as a marker has some added benefit as I find myself leaving high level comments about some of the next scenes (I had nothing planned beyond the start, but the natural progression leads to speculation about future events or details). This is some of that looking ahead data that this project wanted to capture. Perhaps there should be a FUTURE keyword to wrap these things under? It would basically be a THOUGHT for world building ideas, but not specific to the current part of the story/​narrative.

  Anything that goes into writing or crafting needs to be captured in “real time” which means dumping it right in the middle of whatever you are doing.