Nothing is helpful until you need help. The fact that you think AI is helpful for you don’t have to force me to use AI. If you pay me $100k+ per year and you force me to use your tools to do my job why you even hired me ? To force this cult on me or to do the job ?

I don’t need you to use you turn signals, but it’s helpful. I’m pretty sure I could come up with a multitude of situations where your first sentence is refuted without contest.

I thought the turn signals were required by law

In a lot of countries it isn’t.

Or it isn’t enforced, which is about the same thing.

But technically different ( ͡° ͜ʖ ͡°)

Honestly if they hired someone who refuses to use AI in 2026 it’s on them.

It’s pretty normal for an employer to dictate what tools and applications an employee uses. If you prefer gitlab but your employer requires teamcity, you don’t really get a say, right?

However, if you think you can meet your deadlines without using a code generating AI tool, then don’t use it. Ideally, if they are as useless as you imply, no one will be able to tell whether you’re using it. The conflict should only really appear if they do increase productivity, and you refuse to use them, resulting in less productivity than your peers.

The conflict should only really appear if they do increase productivity, and you refuse to use them, resulting in less productivity than your peers.

While it should be easy to point out where this helps, it very much is not in reality because AI accelerates execution on ideas, but corporations nearly always suck much more at deciding which ideas to actually implement in the first place.

I predict that the longer this goes on, the more useless, untested half-features will be stuffed into software and the more bloated “fall back” implementations full of duplicated spaghetti code will exist.

If the tools don’t stick around because it’s too cost prohibitive to keep using them, we’re going to be cleaning up after the bots for a decade.

It’s been a boon for personal slopjects for me. But that’s because I could never find time for execution in the past. Now I just have the (free tier) AI slop up my ideas.

useless half features will be stuffed into software and the more bloated “fall back” implementations full of duplicated spaghetti code will exist.

Which happened without AI, too.

Remember that AI, in whatever form or field, only has to be better than the average human in the field, to be useful to a company project.

Which happened without AI, too.

Sure, but not at this speed. The bots can churn out a mountain of non-sense that does not work faster than people can review it and the bots themselves will lie to you and say “it’s all implemented”.

For people who have clear ideas of what to implement but struggle to find the time, this is a big gift. You can work hand-in-hand with the bot and say “nope, not that, this” until it implements what you had in mind. For aimless corporations with “ideation” meetings who cannot stop coming up with terrible ideas that none of their customers want, it’ll hasten the process of them making their software worse over time.

That’s exactly what I’m seeing at my current company.

That’s a fair point, but it’s not an issue with AI; it’s an issue with corporate culture. Corporate culture can and does change, especially with advances in technology, but there will be a churn period where it’s chaotic and mistakes are made.

I agree with that as well. A lot of the problem with my current company comes straight from the executive leadership level. I’m afraid that the culture will only change when they change (or more likely, the company will eventually just go bust).

They bought the lie that AI will replace all of software engineering, whereas I see the picture more like this:

normaltech.ai/…/why-ai-hasnt-replaced-software-en…

A CEO that thinks that they’re going to develop all of the software, even with Claude code, is nuts. They simply do not have the necessary skill set to run a software project even if 100% of the hands-on programming is done by AI. They are not going to be sitting there telling a machine to move things around on a web app. They are going to believe the bot when it says that “it is implemented”.

In this, I think we more-or-less agree; I just don’t see that as a fault with AI, nor as a reason not to continue to develop/adopt AI.

And maybe your point is what Doctorow was trying to make all along, but he pushed too far in the other direction? Even now, Code Generation is a useful tool, as long as it’s used as only a tool by someone who knows what they’re doing.

And maybe your point is what Doctorow was trying to make all along, but he pushed too far in the other direction? Even now, Code Generation is a useful tool, as long as it’s used as only a tool by someone who knows what they’re doing.

In this piece — which I read via RSS feed a number of weeks ago before it was published in this medium — he is writing about what most peoples’ experience is in a company where they have decided to “do AI”, and what most CEOs think about when they think about “doing AI”.

This usually comes in the form of a company chatbot or “agentic workflow”, and in the vast majority of cases these things are very much not helpful and can actively be annoying because attention is paid to developing these things when more practical and useful features could be developed instead (see: …mataroa.blog/…/i-will-fucking-piledrive-you-if-y…).

In other pieces, he discusses — using what I would call sort of esoteric terms that he often defines inline — people in saner places that get to choose how and why to use AI to assist with their jobs. But that’s not the majority of people’s experiences and that is certainly not what CEOs are often talking about.

I think he’s being somewhat hyperbolic in this particular piece, and maybe that’s due to the amount of “AI is changing everything” or “AI has changed everything” that is appearing in LinkedIn / X posts everywhere, when it’s difficult for most to even name a single thing that the chatbot has spit out that has improved lives, and the vendored chatbot attached to an existing piece of software (which is mostly what people end up doing when they “do AI” at a company) is completely useless.

If I take myself as an example, he’s largely right that I have to play this game with internal development teams that come up with “agentic workflows” and try to get me to use them to do my job when it’s a lot easier and better to just…use Claude code myself.

AI hasn’t “changed everything”. The only thing it’s really done so far is produce a lot of chatbots and agentic workflows that aren’t very useful. It remains to be seen if: (1) it can ever be made to work more efficiently, (2) if you can ever actually run a profitable business building or running a “neocloud”, and (3) what the broader effects on software will be from mass usage of this technology.

The early signals on all of these very real concerns (not to mention others such as environmental, cultural, and economic impact) are not very positive.

Company will be able to tell if you’re using it because companies behind sota models come with monitoring. You will have meeting with hr and your manager when you will be forced to use it even if you meet quotas. They will argument it that maybe you can do more and increase your normal quota with expected ai usage increase in productivity even if you had best productivity in team without ai and your productivity stays the same. Suddenly you will be under performing because you are refusing to use ai so you will be first to fire even if you are best employee. That’s how corporations work. They are monitoring usage because it costs them money. For big corporations they pay upfront to get discounts. Company also make many trainings and meetings to encourage everyone use AI models they pay for but nobody wanted.

Part of the problem is AI isn’t a literal thing with a fixed definition. It’s more of a marketing term that translates into “I want you to buy this thing”. So some things are considered AI that are legitimately useful and others are not and then there’s the debate if wether it’s because it really is AI or if it is and AI just sucks.

In other words, corporate leadership is starting from the premise that AI has (or will) radically change the business, and they’re working backwards from that premise to find the evidence to support this article of faith.

This quote is quite funny in this context, because it’s literally what people do with AI. Half of the frontpage of lemmy is AI. The other half is people saying AI is bad with no realization of the irony, because they don’t see good AI. It’s like CGI. Everybody hates CGI, except they don’t. CGI is everywhere. They just hate bad CGI.

What is AI? Explain as if i just woke up from a coma. We don’t have flying cars, we have dumpster deloreans, but everyone is talking about AI. What is it?

AI is a vague term used colloquially to describe a variety of dissimilar and unrelated technologies each carrying a unique mixture of pros and cons.

So it doesn’t actually mean anything.

The utility of the term “AI” depends on what you’re trying to communicate.

Like most things, meaning depends on context. In the most common context of “news about AI” we’re talking about generative AI: image/video generators, LLMs, and to a lesser degree coding assistants. If you want to know more about what these technologies share in common, look up what a “perceptron” is and how it works.

More generally AI could mean anything from a neural network based approach to problem solving, to a completely deterministic, hand coded heuristic. A pure decision tree could be AI in the case of video game NPCs, for example.

So it’s a new synonym for algorithm.

Not quite. Algorithms are definite, you get the same guaranteed output if you give the same input every time. AI is generally more fuzzy logic.

If it can run on a computer, its an algorithm. Definitionally.

Great, Linux is an algorithm.

It might be accurate, but it’s also vague enough that it’s a pretty useless definition.

I mean, yes. Linux is an algorithm. The programmes called AI are also still algorithms - they’re still deterministic, just too much of a tangled mess, too complex, for a anyone too realistically break it down step by step.

There are some math problems that cannot be described algorithmically, but because of that they’re also non computable. Uh, some thermodynamics and quantum stuff from what I’ve heard.

That’s my point.

I’d say it’s more like a way to describe a certain kind of algorithm. One that adapts in some way to inputs that defy simple classification. Ultimately you’re probably just safe to replace AI in modern news and internet discussion with LLM or CNN.

I agree with that, but others don’t. And that’s the problem. It’s not a fixed definition. It’s marketing.

Since that was a lazy question I decided to be lazy and just copy paste it into Gemini. Enjoy! I have no idea what the point of this is!

Artificial Intelligence isn’t a single technology; it has been a massive umbrella term since the 1950s. Broadly, it means creating computer systems capable of performing tasks that typically require human intelligence.

If LLMs are just one tiny branch on the tree, here is what the rest of the tree looks like:

Machine Learning (ML): This is the engine driving most modern AI. Instead of a human programmer writing strict "if/then" rules, we feed the computer massive amounts of data and let it figure out the rules itself. This is what powers your Netflix recommendations, credit card fraud detection, and the algorithm deciding what you see on social media.

Computer Vision: Teaching computers to "see" and interpret the visual world. This is how self-driving cars identify stop signs versus pedestrians, how your phone unlocks when it sees your face, and how medical software spots anomalies in X-rays faster than human doctors.

Robotics: The physical application of AI. This isn't just mechanical engineering; it is the software that allows a machine to navigate the unpredictable, physical world. This covers everything from the Roomba vacuuming your floor to automated factory arms and those creepy, dog-like robots from Boston Dynamics.

Natural Language Processing (NLP): This is the branch focused on understanding and generating human language. LLMs live here, but so do older, simpler technologies like spellcheck, Google Translate, and the early versions of Siri or Alexa.

Expert Systems & Rule-Based AI: This is the older, "classic" AI. It relies on a massive database of human knowledge programmed as logical rules. When the IBM computer Deep Blue beat the world chess champion in 1997, it wasn't using an LLM; it was using raw computational power to calculate millions of possible moves and their outcomes based on strict rules.

Predictive Analytics & Optimization: The invisible math running the modern world. This is AI used by logistics companies to find the absolute most efficient routes for delivery trucks, or by hedge funds to execute high-frequency stock trades based on market micro-fluctuations.

I asked the question to demonstrate something. You didn’t want to participle in the discussion and that’s fine, but that doesn’t mean the question is “lazy”

I didn’t provide you with an answer to your question?

This response is irrelevent to my post.

LLMs are like a dishwashers. You could wash all the dishes yourself and probably get them cleaner in less time, but it’s useful to let a machine do the work, even I though it’s mandatory inspect every single dish to make sure the machine got them clean enough and re-wash a certain percentage of them. Dishwashers are useful, but the world wouldn’t end if we didn’t have them. They also have the benefit of using less water than when do them yourself, which certainly isn’t the case with LLMs.

In the same line of thinking, this is why I hand wash my car, because the automatic car wash either does a bad job (touchless) or will scratch the paint up all over (brushes). Many people don’t care very much about their cars though, so why put in the effort?

I find unreliable tools to be one of the most infuriating parts of my job, and unfortunately my entire job has effectively pivoted to spending all day babysitting an AI that frequently ignores instructions and can’t learn without a ton of expensive fine-tuning training.

I don’t let AI touch any of my personal code I care about, and with how much of the code at work is written by AI, it’s pretty demoralizing. Why put in any effort designing something if it’s just going to get blown away by a coworker’s agent the next day?

Arguing with the AI over why its code review is wrong is also a whole other part of my day now…

I’m not necessarily suggesting Doctorow is correct with this assertion, but on a macro scale what productivity gains will generate AI provide?

Yes there’s a few skills it’s good at coding, medical research, and image recognition being a few.

The genuine question is: has this improved everyone’s quality of life?

Not yet because people are still learning how to use it, but im literally ten times more organized than i was before i started using AI to automate notetaking planning calendaring all that.

I’m happy for you but isn’t really an answer.

“Not yet” isn’t an answer??

It’s a response but when provided as an answer to the question “is AI a net benefit to society” then it’s a “no”.

Ok so it’s still an answer lmao

I work in government and part of my job is writing memos or documents that nobody will ever read but that some congressional mandate requires me to have on file. I absolutely offload that to AI so I can do my real job.

This is another answer that misunderstands my question. It’s changed your job but has it improved the quality of life of the public served by your office ?

What a grotesque job this is.

I work at a sales-first (read: no accountability) software company, ran by inept nepo-babies and a hostile VC. I frequently have to completely re-do all of the sales engineering work when a new project starts, because they refuse to standardize or talk to each other.

After enough time doing this shit by hand, I took my best examples from previous projects and had an agent “fix the errors based on the conventions I established over here”. Instead of going line by line on these big serialized data formats now, I do about 10min of checking the results and editing down the change log.

It’s wack that I have to re-do someone else’s job still, but at least I can force some version of standardization without killing myself on the tedious bits.

That doesn’t really answer my question. It’s changed your job, but has it improved your customers quality of life?

It’s improved mine. I save hours (that I didn’t really have anyhow) on meaningless tedium. I thought that was pretty obviously implied though?

Maybe it helps to say that this was in no way forced on me. The stuff my employer has done to “adapt AI” is laughable and seemed to be nothing more than lip service to the execs who want to slap that sticker on the product.

The gains for me personally have been strictly using it as a tool in niche situations. I prefer to do things the hard way when it’s for me but I’m not breaking my back for a corporation.

But how do you spend the time you saved?

Either you’re arguing in bad faith by nitpicking for some reason, or you are incredibly dense. Do I need to send you my calendar dude? I just freed up some time, maybe we can hop on a call and I can spell it out for you?

I dont understand. That seems like a perfectly reasonable question.

If gen AI has saved you a heap of time, how do you now use that time that was saved?

However they like, are you their mum?

Hah. ooh good one. What a clever snarky joke you have made. An epic burn if you will. well done.

I see Doctorow’s post has summoned an army of AI dweebs, so I will leave you to your witticisms.

You’re getting direct examples of people its helped, and you keep brushing it off and asking what about everybody else. You’re obviously not going to get a single answer that affects everybody, and why do small successes not matter?

Because they’re not really successes.

No one here is saying AI has reduce their workload so now they spend half their time catching butterflies.

The answers are all saying that AI has automated some mundane task that wasn’t particularly important, and now they can spend their time doing something more interesting.

Is the changes to their role, multiplied by millions of employees, actually better for humanity or society?

In a way, this is what Doctorow is arguing: where is the data showing the productivity gains.

My follow up question is, how are these productivity gains being used, if they exist.

Aren’t you assuming productivity gains will lead to free time? Why? It never has. Most of us work a fixed amount of time and get paid for that time. Not the amount of tasks we finish if that can even be measured.

Salaries have also not increased with increased productivity long before chatbots came online. So they won’t suddenly increase now.

AI is much more just like a new laptop you get from work. The new device is faster, has more RAM and allows you to do more, potentially.

The big question is, if the purchase was worth it. Unlike a new laptop, LLMs are an ongoing cost to users and providers. Nobody except for GPU producers are actually making money right now. Hyperscalers make some of the money back they themselves invested into the companies using their services, but in the end the ROI looks pretty dim for anyone.

So even if there are real productivity gains, you won’t see people relax with their kids more. And even if these gains are significant they may not prevent a huge economic collapse so we all will have plenty of time with our kids in the end anyway.

AI has automated some mundane task that wasn’t particularly important, and now they can spend their time doing something more interesting.

That absolutely sounds like a win to me. Why is productivity gains the only thing that’s important? QoL is what matters. Unless you’re a capitalist overlord.

100%! I don’t understand how this is a hard concept to grasp. I’ve been able to make productivity gains, sure, but who gives a fuck? Even if my productivity is the same, my QoL is still much improved when I’m not buried in mundanity.

The problem here is that the technology is owned by capitalist overlords.

And their goal isn’t improving your QoL. If that was the goal we would already enjoy 4 hour work week long time ago.

Doesn’t matter what the goal is, the context of this particular discussion is that AI is benefiting at least some people by making their lives easier. And one guy who’s trying to argue ‘so what, it only matters if productivity goes up’.

With ownership being as concentrated as it is now (more concentrated than during the guilded robber baron age), naturaly, most of the benefit flows to the biggest owners, and AI seems poised to deepen the wealth inequality further.

In a scenario without the insane wealth concentration, and with a dignified and decent economic floor being guaranteed universally, I would be in favor of any AI that does not burn our planet to a crisp and stink up the neighborhood with the gas turbine exhaust.

So in our world as it is, I have to be agaisnt the AI for the foreseeable future.

The later section addresses this

The other question Suresh implicitly raises is: “How can you reconcile the failure of AI in the enterprise with the individual claims of skilled technologists who insist that AI is helping them do great work?”

OK, but the point still stands if you allow that: It has become impossible to tell managers mesmerised by artificial intelligence that the tools are often not, in fact, helpful.

Sure, although I would phrase it I think a bit more pointedly: “it has become impossible to tell managers mesmerized by artificial intelligence that the tools are not, in fact, capable of what they imagine they are capable of.”

This points, rightly, at it being a failure on the part of the managers - not artificial intelligence. And God knows that’s spot fucking on.

They have uses in certain circumstances. They also have lots of negative aspects (that are broader and more systematic, as far as I can see). I think in a net-benefit sense they are not, in fact, helpful (to an organisation, or to a society).

It is a double edged sword.

If you’re already experienced developer and you use it skillfully, validate produced code it can help.

If you aren’t (and unfortunately everyone thinks they are better than they actually are), it actually can do the opposite, it can help generate a lot of junk code.

Now when working in a team usually majority of people aren’t that great developers typically there might be one or two star developers. The problem is that the other people will still use AI and generate MRs. Reviewing those is extremely time consuming and no one wants to do it. It’s weird to say what’s exactly wrong with those MRs they seem to do things but they seem to do things in a more complicated way.

The thing is that no one wants to review those MRs, everyone is afraid to point AI slop, so the quality of code goes down and even the people who previously were good are getting lost in the code themselves.

It depends on the tool to be honest. If it’s Claude code and you’re a programmer, sure that’s helpful. But often it’s an “agentic workflow engine” that’s been developed in house that’s complete garbage, or a chatbot that doesn’t know anything telling you that it can’t answer your question that was put together by your company. These custom tools are almost always not helpful.

Doctorow knows this as well, he writes about this crap constantly and has broken it down by user type in other pieces.

I have tons of random receipts that come in every month that need to be captured, memoed, categorized and sent off to finance.

I made python to read image/pdf/whatever to text, easy.

I regex the value out, hard and fragile

I try to pull the vendors name out, just not reliable

Category… nearly impossible so i need to make some form of memory system to remember once I do it.

Python renames the file with input if it needs it and provides the meta

OR

python to text because it’s effciient

api call to Ollama and let my old 2070 work it out. (highly reliable)

python renames the file and provides the meta

It’s small, low power, hard to do with code, basically the perfect use case. This is a good use of AI

Go Rewrite FFMPEG in Rust… that’s a bad use of AI.

There’s a lot of people out there using it in ways that cost a lot more than human eyes and have horrible societal and environmental impacts.

There’s a lot of people using it for dictation, document triage and basic scripting where it has great advantages and isn’t making the world a worse place.

His article is at odds with a great deal of his book. I wonder if this isn’t just a grab to get hardcore anti-ai to buy it.