Humans aren’t good at predicting the future, but sometimes you can see a trend that promises something great — like “a genie is granting your wish” great. I think this might be happening right now with one of my genie wishes, maybe yours too.
For me it has to do with the piles of unread books I own. There’s almost nothing I want to do more than plow through stacks of 600-page history and philosophy books, but my efforts are mostly thwarted by the cognitive difficulty I have with processing line upon line of printed text. While I’m reading, my attention veers off at least once or twice per sentence (unless I read aloud, which is slower, hard on the throat, and not always appropriate).
It’s not a small impediment to learning. Not to sound dramatic, but those books represent something I want badly that feels locked away from me, like I’m stuck in the middle act of some frog prince fable. Imagine you loved swimming more than anything, but water happens to cause you horrendous itching.
Audiobooks allow me to spend more time reading (e.g. in the car, at the gym) but lapses of attention still occur frequently. I rewind a lot, but I still miss the context needed to understand the next point. Missed context accumulates until the content is mostly lost on me, then interest crashes completely and I stop.
This happens because books, in any form, are essentially long strings of interdependent sentences, which must be read and understood in order. They operate something like old strings of Christmas lights – miss an important “bulb” and the rest might not work for you at all.
When you skipped a footnote
There are other ways to learn, of course, but they too depend on one’s ability to comprehend long, uninterrupted strings of declarative statements (e.g. recorded lectures), or else they’re expensive and time-consuming (formal education, tutoring), or both.
Books are the traditional go-to for self-directed learning, and I would pay a million dollars for a way to reliably and comfortably get their contents into my head. I envy people for whom reading a book is a straightforward matter. There’s so much I want to learn and study, but processing 500 pages of interdependent sentences is about as easy for me as tying off five thousand balloons while wearing loose rubber dishgloves. Despite this handicap, I’ve certainly read hundreds of books to completion, but I’ve abandoned thousands. Too many missing bulbs.
Me reading Heidegger, page 4
Regardless of whether you suffer this particular bottleneck to self-directed learning, we’re starting to get some new tools that could multiply your current ability to learn.
Over the last few months, I’ve been using A.I. tools, such as Claude or ChatGPT, to learn in a different way. Mostly I get primers on things I’ve always wanted (or suddenly want) to know, such as how does jury duty work, what was Hegel actually talking about, or what do tariffs do and why do people disagree so strongly about them? I can then dig as deeply as I like into the topic, down any strand of inquiry.
The conventional method of intellectual inquiry, for most topics, is to find and read a long sequence of declarative sentences published by someone who apparently knows what you want to know. This means books if you want depth, encyclopedia entries if you want summaries, essays if you want opinions, and lectures if you want lectures.
All of these learning forms, however, depend on your ability to follow, sentence-by-sentence, the thin and winding line upon which the author wants to unspool their knowledge, creating a potential Christmas-lightbulb problem. Every lapse of attention during a given “unspooling” creates another gap in the context for everything to follow, creating a state of ever-disintegrating interest and comprehension. Many of us simply aren’t going to make it through the endeavor — some facts get through, but a working knowledge never crosses over to the new host.
Curiosity level by page 78
This isn’t a huge problem for everyone, but I suspect it is for a massive, untold segment of the population. How many students completely disengage with learning material, at some point, in virtually every subject, because they can’t hold onto an interesting thread long enough? How many people check out of the practice of reading at all, early on in life, because it’s more frustrating than rewarding?
Talking to an A.I. like Claude or ChatGPT allows you to inquire into a topic from right where you are, circumventing the Christmas-bulb effect. You can begin with exactly the aspects of the subject you’re most curious or confused about. What even is jury duty? How do they teach jurors to interpret evidence? Or is that even a part of it? Was that thing I saw on Law & Order the way they really do it?
An A.I. can engage your right at your current level of understanding (or misunderstanding). If you need a definition, or more context, in order to proceed, just ask. If the explanation is too general, you can tell it to get specific. If you need a metaphor, it can provide one (or three or four) immediately. If its language is too technical, or too basic, you can adjust that.
Knows how it works
You can tell an A.I. to answer your question in fifty words, or a thousand. You can ask as many follow-up questions as you need. If it mentions a jury-selection rule you find bizarre, you can ask it to fabricate a debate between two people for and against that rule. You can ask why they don’t just do it this way or that way. You can ask for ten different analogies until you get it. Unlike a human, an A.I. is infinitely patient with you and any trouble you’re having.
After a half hour of free-form inquiry you can come away with a much better understanding of almost any topic – certainly better than what you’d get from virtually any 30-minute lecture or period of assigned reading.
Far more learning could happen in this world if more people could remain interested and attentive to what’s being said. Imagine a world in which 10x, or 100x as much real learning is happening, and across a far greater proportion of the population. That’s a different world.
Deserves a better path
Learning via A.I. interaction is especially powerful for examining your existing beliefs, and understanding why people disagree with you. It’s very hard to do this in open conversation with another human being. Conversation about charged topics is easily distorted by partisan judgments, emotional reactions, and fear of misunderstanding. These factors are massive impediments to learning about and understanding world issues.
You can tell Claude your current opinion about how crime should be dealt with, for example, or when we should intervene in foreign wars, and ask what it thinks you’re overlooking. You can tell it what your take is on the Vietnam War and ask it what Gore Vidal, Jane Fonda, or Douglas MacArthur might have said about that. You can have it write a mini-essay disputing your view, or even have it grill you on your position. There’s nobody to be offended, nobody to accuse you of asking the wrong questions or sounding like one of “them.”
Of course, what an A.I. says to you can be biased, or totally wrong, both factually and morally, but that’s true about humans (and their books) too. All the more reason to seek multiple framings of each question.
[NOTE: Judging by some of the comments I should perhaps emphasize that I’m very aware that AI chatbots are somewhat clunky at doing some of things I’m talking about above, and you should not assume that they’re right about anything. What I’m most excited about is where this tech will be in five or ten years.]
My excitement about this technology is not an indictment of books, not at all. I love books. Books are lindy. I love to sit in a chair and follow someone’s printed, sequential thoughts about a topic, when I can stay on the same wavelength. Books are wonderful, but they don’t serve every attempt to learn.
Great, but not at everything
The future potential for A.I. assisted learning is incredible. We’re about to go from rubbing two sticks together to widespread access to lighters and matches. Say you’re studying for an exam in a dry topic like history or economics. Instead of force-reading a textbook, you can study with a small team of virtual tutors — subject matter experts who can explain to you the vital concepts using language you understand, and analogies related to your actual interests.
Imagine putting on some VR goggles and walking with a virtual Socrates on a shaded stoa, while he expertly leads you, question by question, evening by evening, to a genuine understanding of Hellenic philosophy (or for that matter, American history) at a level strong enough to ace a real professor’s exam.
Philosophy is useless? Tell me, Steven, how do you know that?
I can understand being sketched out by this kind of technology. It seems inevitable that A.I. will change the world profoundly, and quickly. People are going to absorb themselves, sometimes too deeply, in virtual spaces. (Actually this has been happening for decades.) Propaganda will have many new avenues. But what’s better to combat it than a much more knowledgeable population?
I don’t know what’s going to happen. I think it’s a safe bet, though, that one of those profound, A.I.-induced changes will be a massive increase the human capacity for learning and understanding.
For many people, on a personal level, it will feel like the Berlin Wall coming down. We’re about to catch up on a lot of missed opportunities.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
以上版权声明及本许可声明应包含在本软件的所有副本或主要部分中。
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
这些工具大多提供为应用软件生成软件物料清单（Software Bill of Materials, SBOM），即罗列出软件所有依赖项的名字、协议属性和代码数字签名等等。这应当会成为未来软件供应链一个必然的趋势，欧盟今年出台的《网络弹性法案》就有此要求，从制造业的发展来看，物料清单也是产业发展成熟的一个最佳实践。
其实，前段时间 Linux 基金会移除 Linux 项目中特定开发者的 MAINTAINERS 权限，跟上面分析的“开源断供”风险还有一些不同。
首先，根据目前的信息，Linux 基金会实际做的操作是根据一份美国政府的制裁名单，确保被制裁的个人或为被制裁企业提供服务的个人不出现在 MAINTAINER 列表上。这在规则上，并不影响 Linux 项目接收这些人提交的代码补丁。此外，被制裁的个人和企业实际上仍然可以自由地使用 Linux 软件。从供应链角度上看，没有直接的“断供”风险。
但是，这样的行为对特定下游用户的使用仍然是有实际影响的。例如，移除特定 MAINTAINERS 可能会导致 Linux 的某些模块现在事实上处于无人维护的状态，因此，原本将这些模块视为可靠依赖的下游，将不得不重新评估如何处理对相关模块的依赖。此外，企业由于受到制裁，而无法将任何员工培养成关键开源项目的维护者，可能会严重影响企业采用该开源项目的成本评估。这些就是更加复杂而大部分企业暂时遇不到的情况了，本文不做展开讨论。
登录
