sameers

Also implicit in my response was the assumption that college grads don’t generally boo their commencement speakers. Perhaps I am wrong, perhaps that’s pretty routine. And yes, I agree these are just platitudes, that have gotten very tired - but isn’t that how most commencement speakers are? Platitudes, feel-good stories, motivational cliches, all coming from people who lucked their way to the top of some pile and thereby qualified to be invited to graduation? To go back to my earlier assumption - surely the vast majority of commencements don’t lead to booing, in spite of how disconnected and distanced most of the speakers probably are?

I am genuinely trying to understand the nature of the polarized reaction to AI. Just from the perspective of its value to me, I feel buoyed by it, and not worried that it’ll depress my personal value, my financial future, etc. And yes, again, believe me, I get it - I get that I have a lot of cushions and privileges that allow me to feel that way. I am by no means imagining everyone else is in the same position.

But I’m really fascinated at how much more negative the reaction to this technology is in the US, vs. elsewhere [1]. Are Americans somehow more perceptive and alive to AI's danger, than, say, the Germans and Indians that seem to be more willing to find something to be “excited" about, with this technology? Others commenting here have said that the speaker's message, that this technology has just made their college debt much harder to work off, is naturally off-putting. Okay, I confess, I guess I just don’t see LLM AIs as representing such a significantly more dangerous threat than so many other factors that have roiled the American economy over the last 2-3 decades. Offshoring, rising costs, stagnant wages, loss of labor power, inequalities compounded by a severely tiered educational system, constant cuts in public investments in social services - these have been secular trends since the 70s. Have commencement speakers been booed all these years, every time they spout some boilerplate language like, “globalization is the new reality, let's get prepared for it,” or, “Finding your own brand is essential to participating in the 21st century economy”?

I think part of the answer lies in this quote from the NYT article [2]: “Earlier in the speech, Caulfield [the speaker] had lost some of the crowd by praising wealthy corporate leaders, including Jeff Bezos. It didn’t go over well. [A fine arts Bachelor’s graduate that day said,] “Using Walt Disney instead of Bezos would have felt generic to me, yet still demonstrated she understood she was speaking to a crowd of artists.”

There’s an aspect here I think not of opposition to the technology per se but what it says about which large-scale social investments seem to matter to those in power. I think it’s somewhat amusing, and telling, that this person thought to reference Disney of all people - someone that by no stretch of the imagination could be described as a model of corporate responsibility or of support for labor - because in this view, Bezos is so diabolical, even Walt Disney, by virtue of his creative talent, is a benign example of corporate success.

What “Bezos” represents is the idea that the financial and political elites would rather gamble the economy on multi-trillion dollar bubbles, that aggressively extract even more natural resources from our shared commons, or send the privileged few to gaze at this planet from the “edges of space,” than figure out how to keep, say, the cost of insulin low. That’s what got booed, I think, more than the what “AI” represents.

References:

1. https://www.pewresearch.org/global/2025/10/15/how-people-aro... 2. https://archive.ph/UkTaD (NYT)

The relative ease of transitioning to a permanent status meant there was a greater incentive to invest in living in the US, both in material and cultural terms. That may now diminish, and immigrants may remit more of their wealth as a hedge. For immigrants of some countries, that has already been something on the mind now since much before even the first Trump administration. Even transitioning to permanent status has become much harder, leave alone obtaining citizenship.

Maybe this administration's stricter (and scarier style of) enforcement may spur that shift now, but it would have been foolish of immigrants from some countries to have not paid attention at all to the changing conditions over the last 10-15 years.

As a citizen, it saddens me that this administration is reducing incentives for new immigrants to invest here - buy properties, start businesses, start relationships etc. Immigrants will still come to make what they can, and then leave with all that experience and some of those assets, when that could have benefited the country. I am of course hardly the only person noticing how the drive to satisfy the nativist vote is leaving us kicking this gift horse in the mouth.

I did not intend to say that tithing/church donations are equivalent to fraud. I didn't click through to read the story, and therefore hadn't assumed it indicates actual fraud taking place (delayed shipments in and of themselves are unprofessional, but needn't be fraudulent.)

I was implicitly wondering what might motivate people to trust that a brand that has been known to commit fraud, or at the very least to severely under-perform in terms of quality and on-time delivery, can be entrusted with $500 of their money, to produce something the brand isn't known for (consumer electronics.)

Having wondered this, I then proceeded to write my next thought down: Maybe there's more to human motivation than simply the possibility of a material exchange. For example, when one donates to a church, one can reasonably do so without expectation of receiving something tangible and material in return.

I by no means think tithing is fraud, per se. Can and do churches embezzle money? Yes. Does that mean they are inherently fraudulent institutions? Absolutely not. And I am as godless as they come, so I have no desire to defend the motives of any religious institution.

Oh well, I am now at the bottom of this comment pile, so this response, just like other religious acts, is one merely of personal penitence, lol.

If the H1B program had been dominated by say nurses, there is a substitution effect of needing to find and train local (citizen) staff because of the onsite nature of the job. But that's not what IT work is like.

I would like to imagine this motivates them to invest in training native-born workers, but I really can't see that happening, given how they've operated the last 30-40 years.

Umm, yes, yes sir, they do. All "guys," good and bad, planning a war, especially 20th century total war, do exactly this.

The article at least makes it look more like an issue in terms of American law rather than Taiwanese law/policy. I'm curious if a Taiwanese reader would view it as betrayal, or as just corporate fraud.

The assumption there is doing a lot of heavy lifting. You've presented something that's more like a "treatment" for a movie or TV show, rather than anything approximating what's actually happening with LLMs right now, or that can be seen to be happening in the near future based on current trends.

You're right that "[what is not debated is that LLM has changed our industry." Certainly, having a tool that can perform the task of reading all the prior knowledge and documentation and producing simple bits of code that hew to this documentation is transformative. How much someone says LLMs are transforming their workflows depends on how good they were at doing the above task themselves. I've never been good at reading manuals, and reading other people's code - for someone with my brain, having an LLM that can quickly and accurately (enough) answer pointed questions about very carefully constrained problem descriptions is a real boon. It scales me tremendously - projects I had given up on because I couldn't read up and act on the docs for all the different starting pieces have suddenly become something I can imagine at least getting started on.

I still have to plan the project, figure out all the questions at each step to ask about each piece, understand which of the LLM's concerns and objections are valid/relevant enough for me to pursue them, and understand the errors I continue to run into along the way enough to be able to state them clearly back into this "reference finding" mechanism.

The question of "software development" isn't turning into a despairing "What's there left for me to do?", but rather a much more hopeful, "Given that I am past the starting hurdles on so many projects simultaneously, which of these projects should I try to bring to some really interesting point of completion?" Now the constraining limit is more on how well I can conceptualize the underlying problem and less so on how quickly I can recall specific syntaxes for configuring a half-a-dozen different pieces of software.

Simple case in point: I wanted to throw up a web app inside a Docker container in an Ubuntu VPS. I just can't remember all the intermediate steps quickly enough - creating a secure enough deployment user/group setup; making sure Ubuntu is up-to-date; writing the docker-compose.yml file; figuring out what diagnostic commands to run to assess the various errors I run into along the way; working out my DNS and proxy server setup; and so on - even though each one of these is, in the grand scheme of things, a very basic step. Someone who needed to work through this list of steps, who has been creating micro-services using similar workflow templates in their company for 2-3 years, would have had little gain from using an LLM over just doing what to them is muscle memory. For me, it was the difference between, I guess I’ll just have to play with this idea as a toy on my laptop vs. Wow, this is running in the public Internet, with auto deploys from Github, and scheduled jobs running daily web scraping tasks for me while I sleep. And I could get there in maybe an hour’s worth of querying the LLM, reading error messages, and carefully triaging which ones to fix. It’s easy to underestimate just how much domain knowledge is still necessary even in this situation.

To go back to your analogy - now that you have a robot in your restaurant that can be fairly easily tasked by you to, say, make the perfect dough for a simple enough recipe; re-arrange your tables to add 3 more tops; be instructed to call all the butchers in town asking if they have specific cuts of meat; and so on and so forth - what can you, a star restaurateur, chef, food entrepeneur, do more of? It doesn’t mean that running that restaurant suddenly became as easy as the simplest of its composing tasks. The robot is flipping the burgers perfectly now so you can maybe save money hiring someone for that specific task. But the mere act of perfectly flipping burgers still leaves you many, many steps away from satisfying even one evening’s worth of diners.

All creative mechanical processes - whether it’s factory made furniture, reproduced prints of the Old Masters, or now gen AI outputs - rely on some examination of prior human output, because the judgment of what people will want to use or consume has already been made historically, and the mechanical processes are different ways of distilling those decisions into something that machines can reproduce.

In this particular case, the restaurant owner put effort into writing a prompt and presumably reviewing many possible outputs to pick one. What has happened with gen AI is that anyone who can somewhat visualize the artistic product they are looking for - a blurb in a specific style, a picture that has certain visual aspects, sounds that match some set of sonic characteristics they like - can now produce something passable without involving another human being. That has happened many times since the start of the Industrial Age - a vacuum cleaner does a passable job of cleaning surfaces, microwaves do a passable job of cooking foods, synthesizers do a passable job of sounding like a human orchestra, and so on. All these tasks could only be done, at any level of quality, by humans - now, humans are still capable of being more innovative and detail oriented than machines can with a variety of cognitive tasks, but machines have gotten to the “vacuum cleaner” level of producing equivalent outputs.

I get the anxieties and fears coming from seeing a whole class of aesthetic and cognitive tasks being taken over by machines, because the ability to compose a marketing slogan feels closer to the notion of human identity and cultural know-how, than knowing how to clean a home well. Is that what explains the incredible degree of paranoia and resentment we are seeing, evidenced in this case by people wanting to hurt someone’s business/livelihood because of how mad they are?

edit: forgot what I intended as a "reference link": [1] https://duckduckgo.com/?t=ffab&q=llm+ai+transformative+use+c...

Re this being the FAA's choice, I was reacting to this line in the reporting: "On April 10, Levine and his lawyers pressed ahead by filing an emergency motion [... which... ] may have expedited the government’s next move [to replace] the sweeping flight restrictions with a “national security advisory” [and dropping] all mentions of flight restrictions and criminal charges." Maybe Ars is being too rosy-viewed about the causality there, idk. I have no partic feeling one way or the other though I do want to take whatever comfort I can in the notion that the "system of checks and balances" is working. I'd rather go to bed thinking it is, than tell myself cynically that this was just another whim of an agency, with no real principled attitude.

I believe that the Trump administration in particular - not Republicans as opposed to Democrats - has abused agency independence in a manner unprecedented in recent American politics. I think agencies SHOULD act autonomously to determine specifics just like this one - what vehicles/devices, with what capabilities, can fly where and in what manner, and that we SHOULD value "expert advice" in such situations instead of using that phrase as invective. I think the American people should celebrate that we grant such freedoms because it lets us all benefit from expertise - but they should also understand that there is a price to pay in vigilance, of having to challenge the legality of agency actions if the particular implementation of regulations infringes on constitutional rights. But it's not just litigation that will prevent abuse - the first line of defense against it should be the expectation that administrations will consider themselves beholden to certain social norms of cautious use of power. Do you believe that there is no daylight between this administration and previous ones in terms of how they view what norms they ought to consider themselves bound by? That's a genuine question, not a rhetorical one - if you don't believe that, I am curious to know more.

I don't think legislatures can possibly identify a priori all the ways in which rights could possibly be infringed and make their grants so granular that agencies can't possibly find abusive interpretations. Those can only be determined in specific, real, cases, when fallible individuals attempt to meet the legislated objectives by taking concrete action. I don't understand this idea that federal agencies have become "unaccountable" merely because they issue intepretations every day as and when they encounter real-world situations. The Chevron doctrine seemed a perfectly fine compromise to me - how this court thinks the legislative body can magically divine all the future possibilities and encode them into the acts that govern the agencies is just beyond me.

edit: nm, archive can't get past the JS block.

I think what you are facing is a sense of how much the tools are just that - tools, and no substitute at all for what an experienced developer would try to plan for. So yes, I am sure everyone who's using AI tools for the first time has made this "newbie" mistake.

BUT I have also heard that the tools come with "modes" (or agents, or skills, or whatever you want to call them) where you can have it act _like_ an architect, that points out the things you ought to ALSO do. Like write tests, ensure idempotency etc.

I am curious what your experience would be if you went through some cycles of having the AI simply review its work and suggest improvements. Clearly you've already learned some things that you ought to add - I think documenting the progress you make as you discover these things would be immensely helpful to others like you who are new to this!

Good luck :)