TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Why the future of AI is neurosymbolic. (A rare optimistic post from Gary Marcus)

36 点作者 garymarcus10 个月前

11 条评论

Animats10 个月前
So we need to do more Good Old Fashioned AI, and get off my lawn.<p>He has a good point that model-less LLMs have serious trouble with problems that require a model. But predicate calculus hasn&#x27;t worked out well as that model.<p>Many years ago, I took John McCarthy&#x27;s &quot;Epistemological problems in artificial intelligence&quot; class. He laid out the missionary and cannibals problem informally, then wrote it up in a predicate-calculus notation, and then applied his &quot;circumscription logic&quot;[1]. As he wrote it up in his formulation, I thought &quot;and then a miracle occurs&quot;.[2] As with most story problems, it&#x27;s getting the problem into the correct formalism that&#x27;s hard. Turning the crank on the formalism is usually straightforward.<p>Much of the AI community spent the 1980s beating on this problem. A large number of very smart people tried to solve it. Many things were tried that were less rigid than predicate calculus - probabilistic logic, Markov chains, fuzzy logic, etc. All mostly failed. The AI Winter followed.<p>The classic critique in this area is &quot;Artificial Intelligence meets Natural Stupidity&quot;, by Drew McDermott.[3] That&#x27;s from 1976, and still relevant to this argument.<p>LLMs, though, might be able to use such models. Something to try: put a story problem into an LLM, and ask it what formal methods might help solve this problem. Then ask it to convert the problem into each of those formal methods. Then use something like Mathematica on each formal method. LLMs can&#x27;t do logic problems, but they can sort of write code and translate between languages. So maybe they can do the miracle part. Anybody working on this?<p>[1] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Circumscription_(logic)" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Circumscription_(logic)</a><p>[2] <a href="https:&#x2F;&#x2F;trevor-hopkins.com&#x2F;fiction&#x2F;miracle2.jpg" rel="nofollow">https:&#x2F;&#x2F;trevor-hopkins.com&#x2F;fiction&#x2F;miracle2.jpg</a><p>[3] file:&#x2F;&#x2F;&#x2F;home&#x2F;john&#x2F;Downloads&#x2F;Artificial_Intelligence_meets_natural_stupidity-2.pdf
评论 #41097100 未加载
评论 #41095766 未加载
评论 #41095519 未加载
_nalply10 个月前
Perhaps a good way would be to define a logics language for AIs or use an existing one like Prolog and let the LLM generate code then run it. It&#x27;s a variant of giving LLMs access to some system and let them iterate till they find the solution.<p>The idea: When programming sometimes the solution is not exactly right but with feedback a better solution can be found. I once made that explicit to the LLM and I played the evaluator for the LLM and it seemed to work better. I am not surprised. Even humans usually don&#x27;t just program on paper. In the early days of the computer science they had to do that. I experienced that myself: my first BASIC program was on paper and then when I had access, I typed it in. The experience is bad. I had to guess without feedback.<p>I can imagine that it doesn&#x27;t solve all problems because not all is solvable by trying out things or the formulation of the problem doesn&#x27;t describe the problem correctly. But it is better than not trying out things at all.<p>Please see Animats comment: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41095417">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41095417</a> and the answers there. They are saying more or less the same thing but more detailed and knowledgeable than me.
评论 #41098876 未加载
tim33310 个月前
&gt;the future of AI is neurosymbolic<p>There are a number of ways the future could go but comparing language models like chatGPT to human thinking the more obvious way forward might be visual reasoning and spatial modeling.<p>A lot of human thinking is vision related as in &#x27;I see what you mean&#x27; &#x27;picture this&#x27; and so on. Also comparing current AI to humans it&#x27;s getting quite good at written exams but terrible at physical stuff like getting some milk from the shop and making a cup of coffee. Also the emergence of something like physics in the likes of SORA suggests it&#x27;s possible.<p>On the other hand symbolic logic along the lines of algebra is quite a specialist area that humans have to be taught in maths classes and many get by without learning. I presume AI will get good at it but it doesn&#x27;t seem the most obvious way forward unless you want it to do maths.<p>(by the way I came across a paper on trying to go this way <a href="https:&#x2F;&#x2F;spatial-vlm.github.io&#x2F;" rel="nofollow">https:&#x2F;&#x2F;spatial-vlm.github.io&#x2F;</a> - adding 3d data to a multimodal large language model)<p>I guess human reasoning has three major areas - language, visual&#x2F;spatial and also thinking of other thinking entities like other people, the cat wants to go out etc. Likely consciousness relates to the last category - does he feel like I feel and the like.
评论 #41100567 未加载
tim33310 个月前
I tried the goat puzzle on chatgpt it it got it right with reasoning as below. This suggests to me the systems are improving and don&#x27;t seem at an obvious plateau yet.<p>&gt;To solve this puzzle, it&#x27;s important to make certain assumptions and understand the conditions. Since you mentioned only a man and a goat and didn&#x27;t specify any constraints, I will assume the following:<p>1 The boat can hold both the man and the goat at the same time.<p>2 There are no restrictions on how the man and goat can travel in the boat.<p>3 The goal is simply to get both the man and the goat across the river.<p>Given these assumptions, the solution is straightforward:<p>The man and the goat get into the boat. They both cross the river together. Therefore, both the man and the goat can successfully cross the river by traveling together in the boat.
评论 #41096768 未加载
fishermanbill10 个月前
Not sure this really holds: there is no symbolic LEAN type system in any animal brain as far as I can tell. I do think the point about the AI needs to sanity check is right. I imagine all this is will fall out from the visual, audio and other senses feedback loops though. I dont see the fundamentals needing explicit algebraic reasoning - that will surely come later.
setracer10 个月前
The brain doesn&#x27;t have a CPU-like architecture doing symbolic reasoning. In fact, we now offload this kind of reasoning to computers because they do. If you want to create human-level AI, this is not the way. You&#x27;ll be able to create really powerful and useful niche systems like Deepmind has, but not a truly general reasoning machine.
评论 #41097163 未加载
评论 #41096347 未加载
judiisis10 个月前
He is soliciting upvotes in violation of hn guidelines <a href="https:&#x2F;&#x2F;x.com&#x2F;GaryMarcus&#x2F;status&#x2F;1817628149381054643" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;GaryMarcus&#x2F;status&#x2F;1817628149381054643</a>
fsndz10 个月前
Some people even argue that we won&#x27;t be able to build any AGI... <a href="https:&#x2F;&#x2F;www.lycee.ai&#x2F;blog&#x2F;why-no-agi-openai" rel="nofollow">https:&#x2F;&#x2F;www.lycee.ai&#x2F;blog&#x2F;why-no-agi-openai</a>
评论 #41099373 未加载
orbital-decay10 个月前
This keeps emerging again an again, and the answers are pretty generic.<p>1. <i>Large language models</i> as a concept are not going anywhere anytime soon for any reason. Simply because there&#x27;s no other source with a huge slice of human psyche encoded into it than the language itself and the corpus of texts in it. Humanity collectively did a massive amount of gradient descent on the language over generations, and it will stay as the primary source. That doesn&#x27;t mean that other sources don&#x27;t exist, of course.<p>2. Dataset quality matters at least as much as the architecture. There&#x27;s plenty of low-hanging fruit available in preprocessing the data and &quot;textbooks for models&quot;. You learn to count in a decimal system from both memorizing the number sequence and the explanation of the algorithm, not just by looking at millions of examples! There&#x27;s plenty of a bit higher-hanging fruit available in hardware improvements and optimizations.<p>3. Calling a transformer a token predictor, stochastic parrot, autocomplete on steroids, etc. is of course right but kind of misses the point, like calling human brain a nerve impulse predictor (and the brain also has no &quot;inherent way of verifying whether their predictions are correct&quot;, using the definition from the article). Reasoning about this in ill-defined terms like &quot;understanding&quot; or &quot;knowledge&quot; or &quot;intelligence&quot; is not useful at all. There are many differences between humans and LLMs, but the most high-level one is that humans are autonomous agents that exist in continuous time, and transformer&#x27;s lifetime is the time required to compute a single token. Repeat the process for multiple tokens and you have something more complex. Add an external loopback, and you have a chatbot with memory, partly capable of doing things unexpected of a &quot;word predictor&quot;. Make the loopback more complex, and you suddenly have an... autonomous system that exists in continuous time. Sure, it&#x27;s <i>extremely</i> crude and primitive, and that loopback probably also needs to be replaced by something way more advanced in the future, and, and, and, and...<p>4. Reasoning and symbolic computation comparable to human abilities (which are also pretty spotty and error-prone) might or might not emerge as a result of scale and simple loopback mechanisms in models. You might or might not need an external symbolic engine as the author says, or maybe you can reduce it to another model of a different type, or maybe it&#x27;s all wrong. Current models are still orders of magnitude smaller and simpler than the human nervous system, and plenty of things in LLMs already changed by simply increasing the scale.<p>5. Other than all of the above, sure - transformers or another flavor-of-the-year architecture might give way to more advanced ones. But the basic principles will remain, and language models are not going anywhere.
评论 #41096422 未加载
artninja198810 个月前
To me neurosymbolic seems to be a mostly tribal distinction. Also Gary Marcus is just trying to make Google&#x27;s achievement all about himself again because he&#x27;s a narcissist. But what else is new
评论 #41095374 未加载
评论 #41095342 未加载
godelski10 个月前
I know this is a contentious subject and tbh I&#x27;m not a big fan of Gary myself, but I&#x27;d like to make a simple explanation as to why AGI requires symbolic manipulation. Note that I&#x27;m careful here and not saying it needs to completely neurosymbolic nor am I saying the advancements we&#x27;ve made won&#x27;t contribute to AGI (they will). Unlike Gary I am not willing to ignore the advancements that LLMs have made and the utility that they provide. Their limitations do not invalidate the improvements. But this is the identical error those make that dismiss symbolics as an avenue forward.<p>At the end of the day, the current framework of ML operates by ingesting data, modeling that data, then iterating on that model to better fit data[0]. This is something humans and every living creature does. But many animals, humans included, do so much more. The problem comes when we get to understanding abstraction, and an issue is that we operate at such extreme levels of abstraction that it is easy to miss. After all, we were designed to be better at recognizing differences because it allows energy savings for business as usual settings. The problem is that pattern recognition leads to no such viable path for our levels of abstraction. It may not be obvious, but we are symbolic manipulators. Our language is composed of symbolic manipulation, our code, and our math. The last of which may be the most clear example of the distinction. But this might be a bit uncommon to see unless you&#x27;re from a strong science background.<p>We all know that there are great differences in numeric&#x2F;empirical solutions from analytic solutions. The latter of which is considered both harder and much more rigorous. The latter is naturally causal and interpretable. The reason we do numerics and empirics is because limitations. But analytics is why a physicist can sit in a room with a pen and paper and (eventually) discover fundamental laws of nature. Many of these achievements are not solvable by observation alone[1]. But these equations are symbolic. The symbols are the abstraction. A major advantage of the symbolism here is that once we are able to formulate solutions and the rules of the symbolic system, we can manipulate as we please. This has so much more flexibility than a numeric solution. This is the underlying reason the theorist exists! It allows for us to quickly and accurately ask new questions and find errors or limitations. This kind of manipulation allows us to ask why gravity is an inverse square law and to understand why it is exactly 2[2] and not close to 2. It allows us to set concepts aside that we might call a constant (when the resultant is unitless), solve currently tractable factors, and even then later determine what this constant is (often the job of an experimentalist). We may even later ask ourselves how a constant may be decomposed into other factors. The symbolic nature allows us to pattern match in ways we wouldn&#x27;t be able to with numerics. There is just a high level of abstraction that we are unable to do with the numerics. Abstraction that we rely on to create and understand the world as it is.<p>So the great question in AI&#x2F;ML is not if these systems need to do symbolic manipulation. It is if a machine can learn to do symbolic manipulation through numerics. This is still unknown and there are arguments on both sides (right now the case is stronger against this happening through data processing). The only naive thing would be to not pursue both paths (well there are many paths). We&#x27;re venturing into the unknown. We&#x27;ve gotten a long way through the methods we&#x27;ve been using and this is reason to continue down that path. But at the same time, this is not reason to pursue others. We&#x27;ve never seen that happen in the past. All technologies undergo radical shifts that are no apparent to those on the outside. Imagine if we didn&#x27;t pursue LiPo because lead-acid was working. If we didn&#x27;t pursue transistors because vacuum tubes were working. LEDs because incandescent. And all the new technologies began as worse (often much worse) than those they later replace[3].<p>Lastly, I want to speak to investors directly. If your goal is to invest in a new company that will make AGI, you are likely to lose if that company is pursuing via LLMs[4]. There are already major players in this space that are far ahead and have more momentum and funding. There are things that they are missing that others might see, but they have the capacity to find those limits and fix them[5]. Instead, you have a better chance on what appears riskier: less developed avenues that also have explainable avenues towards the goal. It is high risk, but it always was. Here&#x27;s the thing, I said the question of numerics leading to symbolics is still open, but another way of looking at this is that we know symbolics is necessary (or at least we know symbolics is sufficient for intelligence).<p>[0] Note the dependence on the previous estimate&#x2F;model.<p>[1] I want to note that there is a feedback mechanism which is what I reference in my first paragraph. The theoretical physicist stands on the back of experimental physicists just as the experimental physicist stands on the back of the theorist. An untest{ed,able} hypothesis is no theory and a fitting data is not a physical model without theory. See Fermi&#x2F;Dyson&#x27;s conversation about fitting an elephant.<p>[2] This is a calculation you will do in an upper division classical mechanics course (physics).<p>[3] Often also with people questioning why we should pursue these other paths, not recognizing -- or unwilling to -- the limitations of the current technology. And no technology is without limits. That alone should be reason to pursue other avenues.<p>[4] If you&#x27;re investing in products, then pursue LLMs. They are much more mature and you have the infrastructure of research behind you. You can also likely adapt to a changing underlying technology.<p>[5] Unless your real goal is acquisition by said players