How to play: Some comments in this thread were written by AI. Read through and click flag as AI on any comment you think is fake. When you're done, hit reveal at the bottom to see your score.got it
A new and growing fear I have is that AI cuts off some of the most well-trodden pathways to intellectual growth. Every time the AI solves a problem that I would not have been able to efficiently solve alone, it has replaced an interaction that previously would have happened with a mentor/supervisor/code reviewer. This is faster for the person seeking the micro-assistance, but there is a flipside: fewer human-to-human acts of mentorship and shared problem solving. We lose a lively ingredient of team formation, expert formation, and a source of joy for all involved. As this scales up it feels plausible that we end up with a wider deagradation of intellectual standards. I like the Cognitive Comons framing.
Ran a longitudinal cohort on exactly this a few years back, before LLMs were the trigger. Cognitive offloading isn't binary. GPS use measurably degrades spatial memory over time, but calculators didn't kill arithmetic intuition for people who still had to estimate. Depends what gets outsourced and how often you still practice the raw skill.
Mentorship time was already scarce before AI showed up — most people never had a mentor patiently walking them through problems. AI didn't replace that relationship, it filled a vacuum that understaffed teams and busy seniors left wide open.
For anyone that does LLM supported work it's crystal clear that skill decline is real. Surprisingly, the leadership at my work doesn't give rats ass about it. They published AI engineering manifesto, pushing for loops, workflows, advocating for producing many times more code. All of this ignoring the elephant in the room.
Cynically they may be banking (however misguided it turns out to be) on model skill in aggregate growing faster than the skills of their staff decline at which point, bin the staff, use the model.
The hype from non-technical people about AI would make you think they already discovered AGI.
I don’t think it’s as clear cut as you make it out to be.
Maintenance <> growth
Once you got a skill it’s quite hard to actually lose it. It’s like riding a bike as they say.
I find all the reading of those walls of code exercises my mind better than the writing did. I’m sure I lost memories of how to read a file by hand or balance a tree but who cares?
I am thinking about architecture and design a lot more these days and every decision I make has to be actually argued for even to myself. I can no longer lean back and say “that’s how we always do it” or “too expensive to change now” as I find many people in practice actually do. They were just coasting on premade architectural choices and their “skills” consisted of knowing arcane incantations and syntactic details completely unrelated to the (business) problem at hand.
I am not convinced many developers actually have the skills they think they have. They could wrestle syntax and mess around with tooling, but could they abstract properly? Define clear semantic boundaries? Have proper civil discussions about responsibilities and where they should lie on the right level of abstraction? Nothing has changed in that regard. If anything that part has been amplified. (“taste”)
It doesn't seem clear to me at all. e.g. I've been having a fun time learning Godot this last week with codex laying down groundwork for logic, generating assets, etc. This gives me something concrete that I can work with and ground myself within instead of starting from a blank canvas. I don't much care for the actual scripting logic because it's all basically trivial to me with a few decades of programming experience, and I'm not going to be able to develop my artistic abilities to a useful capacity anytime within the next few years, so I can instead focus on learning architecture/organization patterns and game systems that I'm interested in while I read through docs.
So basically, if you use it to wave away things you've mastered or explicitly don't want to learn at the moment, you can stay focused on things you are figuring out. I don't see it as any different from writing "by induction" in a math proof without writing all the details because you and the reader know you could easily work them out. In this way I can learn things that I simply would not have the time to dig into before, increasing my skillset. As is was before AI, metacognition is the most widely useful skill to have.
I don't see it being "crystal clear" that skill decline is real. People are doing work they wouldn't have been able to do before. They are getting a more broad exposure to technologies than they have before.
Similar to the argument made by John Blow several years ago. Though, Blow blamed frameworks/engines and layers of abstractions. I wonder if the authors came to this framing themselves, or listened him.
My approach is to write a bit of totally AI-free code every day. So after a day of Claude, I'll spend at least 30 mins wrestling with something. The gnarlier the better, e.g.leet code or Project Euler-type stuff. I think of it as like lifting weights for the mind. The more I struggle at the edge of my knowledge and skill the better.
The other thing is to give your agent a skill not to solve certain key problems unless explicitly prompted. Write the scaffolding sure, but leave the juicy parts alone. And if I get stuck, I have it enter into a dialogue with me, nudging me towards understanding.
I solved a Leetcode Medium on my first try in about 10 minutes the other day and it felt so amazing. Thinking through architecture, project composition, composability, etc when working with agents felt fine for a while but it made me feel disconnected from what I'm actually building.
Actually engaging my brain to solve the lower-level, on-the-ground code allows me to think of better ways to do things while I'm writing them. It's like writing anything. You start with something you want to convey, a thesis, and then it evolves and becomes better as you write it. An agent will just write it with no thought, as in it will reflect one of the LLM 'ghosts' as Andrej Karpathy puts it, doing something in the same way that someone in the training data has done it on a similar or different problem. This is why I get conniptions now when I am sent generated text or am expected to read it on a public forum. It's disrespectful of the time of every person expected to read it.
That said, I wouldn't like to go back to the before times without having the agentic option available. Ideally, businesses should not mandate how LLM's are to be used at their company, and just let the devs find their own flow. That is, if quality is even a factor that any company optimizes for anymore.
We will look back on this era as one of transition. AI is in its "look monkey can do tricks how cute" era - but it won't be long. For instance the era of "whoa" is happening right now in mathematics (Anthropic researchers are pushing math's frontier with little more than the prompt "you can do it" - ie. brute force), and in coding agents we're crossing over - look into leading-edge benchmarks like SlopCodeBench that are pushing labs to RL for long term codebase health and not just problem-solving. The reason I say coding agents will go in this direction is labs are competing to have the most appealing models and so seeking out new unsaturated benchmarks they can hill-climb and show flashy results from, demonstrate to customers they're the best and deserve spend.
What these cognitive tools do is make the mundane work we're lamenting the loss of redundant. Who cares if junior devs can't code? When this stuff really gets going they'll be using their answer boxes to both code AND to have have design discussions with a form of intelligence that has every PhD ever obtained and infinite patience.
Consider: long division used to be in school curricula; our forebears had to know it to be considered "educated"; and yet I, in my 40s, was never taught long division.
The pseudo-academic style of the article is awful. All abstractions and buzzwords, no examples. Also, the tragedy of the commons concept is far older than 1968. The term "common" refers to common grazing land upon which anyone could graze animals.
That's an English term. The American equivalent is "open range".
Getting past that, the author has a point. There's a loss of shared expertise when there aren't people around learning and doing something. In the US, we've seen this in manufacturing. The number of Americans who know how to set up a good production plant is much lower than it was in the 1980s. That's the consequence of the hollowing out of American manufacturing. The author talks about AI vs. white collar work, but fails to make the connection with outsourcing vs. blue collar work.
The loss of this expertise has recently been made very clear in the US as attempts are made to scale up weapons production for the US's various wars. Progress is very slow, as has been seen with both artillery ammo and air-defense missiles.
> Also, the tragedy of the commons concept is far older than 1968. The term "common" refers to common grazing land upon which anyone could graze animals. That's an English term. The American equivalent is "open range".
The term is not "common" or "commons" but "tragedy of the commons" which dates back to a book in 1968 by Garrett Hardin. Elinor Ostrom subsequently showed that most or all of Hardin's assumptions and claims were wrong, or at the very least, far from universal, and that the process he describes elides what actually drives the destructions of resources held in common: greed and power.
Must be a glitch in the matrix or I got kicked into an alternate timeline, because until yesterday I never saw anyone here question the general meaning of "tragedy of the commons" and its stated implications, and suddenly it's the second time I see some 1968 book reference brought up and implications that "tragedy of the commons" is debunked, and that it's saying basically the opposite of what I know it to be saying...
I think you might be thinking about Elinor Ostrom and she didn't show that his assumptions and claims were wrong exactly. She expanded on the work to clarify the difference between managed vs unmanaged commons. Showing that successful commons are managed in some way with established rules and limits around the usage. The term still has use especially if it is used to clarify the need for management of common resources to avoid the tragedy.
Us manufacturing is larger than ever and still growing. It is mostly automated though so Tht number of people who work in manufacturing is shrinking. Still the expertise is still there and they do hire people all the time.
I thought the article was going to be about countless millions of minds trying to win the internet lottery and merely reinventing the wheel or failing.
I'm not worried about AI taking jobs. I'm worried that humanity has lost the ability to share at such a monumental level that basic sustenance and financial security are out of reach, even with AI.
After lifetimes of negative reinforcement, the only salvation seems to be the disruption of capitalism itself. Somewhat ironically, the wealthiest and most powerful people in the world seem to be investing trillions of dollars into AI to do just exactly that.
They are not out of reach - people just increase their expectations to ensure that they never achieve satisfaction. I think that's fine but it is optional - if you want to just exist without striving for anything, nobody's stopping you. It's easy to be self-sufficient if you'll tolerate pre-industrial standards of living. There are others who want that too and you can coordinate with them and share among yourselves. This already happens. Modern standards of living and your high expectations are a consequence of capitalism. Ironically, you can actually have your cake and eat it too. So many other people love to participate in capitalism, that they provide enough good things to make self-sufficient living easier today than it was in pre-industrial times.
No, humanity hasn't lost the ability to share - it's the exact opposite - trade (and capitalism) has enabled sharing far beyond what it ever was before. Even better, it's done though voluntary and mutually beneficial exchange rather than people being forced into a fixed role in life which was often something horrible like servant, slave-girl, or warrior.
Perhaps, but I find when I use AI (for fixing things), I learn a few things here and there as the AI-provided answers are often wrong. And when I work around these errors, the learning occurs.
And if the solution turns out to be "we don't need those skills anymore", what will happen to those subsidized people? Better just let them find work wherever they see a demand. Software has gutted many other industries of competent people. Maybe now they can return and lift up the rest of the economy.
This paper misrepresent fundamental concepts from Ostrom’s work, which won a Nobel, and fails to contextualize Hardin’s theories. As a 'human resource development' paper it somehow pretends that capitalism doesn't exist?
Hardin coined "The Tragedy of the Commons" in 1968, elaborated on his theory in his 1974 paper "Lifeboat Ethics: the Case Against Helping the Poor". He was a eugenicist and specifically targeted refugees. He lobbied US Congress in opposition to international famine relief. His evidence-free theories must be contextualized within his overall white-nationalist project. https://www.splcenter.org/resources/extremist-files/garrett-...
The paper asserts that "expertise within a profession" is a commons. As an industry we've never been able to define the specific, task-level role boundaries between PM/designer/engineer across companies. So how is a profession defined? What are the units of expertise here? This paper fails to provide a defensible definition of their commons. Ostrom's commons need clear boundaries and resource units. https://en.wikipedia.org/wiki/Elinor_Ostrom
From the abstract: "*The paper reframes expertise development as collective stewardship*" which ignores capitalism.
Firms are incentivized to build proprietary expertise, eg Slang at Goldman. Individuals are incentivized to build proprietary expertise and use it as leverage for higher compensation. Even if we fudge commons into a more generic collective action problem, this paper doesn't reckon with the tension between market incentives and collective expertise. This tension comes up again and again, in open source, in academia, and corporate L&D programs.
The author goes on to use professional organizations as an example of governance for their Cognitive Commons. The examples of medicine, law, and engineering are particularly bad as these are credentialed, legally enforced enclosures of expertise, and not self-policing or democratic. They are the opposite of commons.
The author should have just written about expertise as a resource pool and avoided the commons. This was a frustrating read. The paper is a disservice to the commons literature.
Minor nit: bar exams and admission aren't self-policing, but discipline after that basically is, state bars rarely disbar anyone. Which kind of proves the tragedy-of-the-commons point about self-regulating professions more than it undercuts it.
We can view the bar association as a self-policing group to some extent. But the unauthorized practice of law is a crime in most jurisdictions, so it's policed by the State. In the context of Ostrom's commons this means a hypothetical 'lawyer-commmons' exists at the whim of the State and is unlikely to survive. In practice, a fake lawyer serving jail time can't get out of jail by negotiating with the bar association for access to lawyering. It's similar for medicine and engineering. I take your overall point that professional organizations have internal rules, decorum, membership criteria, etc. If you're at all interested, Ostrom's 8 rules for managing the commons are worth a read.
Self-policing bar associations are an interesting case for Ostrom's design principles actually, since bar discipline mostly runs on peer reporting and reputational cost rather than state enforcement. The UPL statutes are more a backstop for outsiders than the mechanism keeping lawyers themselves in line.
for a lot of the article i was mentally replacing "ai" with "calculator" and going "yeah we will just need longer education times. more schooling etc."
but the phrasing of "who pays for the increased schooling times?" is a good one.
i think "debt" can be a decent way to conceptualize the cost and repayment of training someone. feels evil to say, but viewing people as firms you can invest in and expect returns upon. you know not all loans will be repayed, but hopefully they'll average to a profit. (risk management etc.)
student loans are. a decent example. the government/private enterprise gives money to pay for education, then this is repayed, providing a financial incentive for paying for someone else's longer schooling timelines. firms investing in training can be viewed as an extension of student loans. but then ah, there are countless stories of how debtor/creditor relationships can be exploited. indentured servitude etc. there are a lot of complications coming to mind. also "altruistic" people who give without expectation of repayment. or the divide between like, communal vs individualistic cultures. (individualism, i argue, encourages the formalization of debt, as opposed to a more communal culture where the expectation of repayment is informal.) you could do math on how many people pay vs how many people benefit, who is the biggest stakeholder, etc.
but i am on my lunch break and need to get back to my work. good article tho. good topic to bring up.
Same fight as calculators vs slide rules, IDE autocomplete vs memorizing APIs, compilers vs hand-tuned assembly. Every abstraction kills some skill underneath it. Commons only tragic if nobody keeps the old muscle anywhere. Somebody always does, out of spite or paycheck.
