I use Claude and Thaura for search in my daily life.
I use Claude as a rubber duck for coding. I also let it review my code for errors or make recommendations. On occasion, I'll let CoPilot or Claude add a few lines of code directly. I don't "vibe code".
I enjoy both uses. I don't feel guilty about either.
This entry was edited (Wednesday, March 25, 2026, 5:57β―PM)
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I don't agree with many of the arguments against LLMs I see on the Fediverse.
LLM usage is not a GHG-intensive activity, especially compared to other everyday activities.
It disturbs me when people imply that LLMs are primarily responsible for climate change. This is not remotely true, and it ignores our real responsibility for working on the climate emergency.
It hurts my heart that smart people I know and respect don't know the main sources of GHGs.
cosocial.ca/@evan/115580076628β¦
Evan Prodromou (@evan@cosocial.ca)
Evan Prodromou (CoSocial)Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I don't use the term "stealing" for ideas, and I don't think it applies for LLM output.
I think that training LLMs on content to produce a model, then using that model to generate new content, is more akin to human learning than to verbatim copying.
I don't think that LLM output is a "derivative work" of the training data.
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Dr Kim Foale
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Dr Kim Foale
in reply to Dr Kim Foale • • •aeva
in reply to Dr Kim Foale • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Aljoscha Rittner (beandev)
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Hmm, I understand many of your arguments and experiences, but I see a break with low-code/no-code platforms. That's because I don't know of any professional solution where these platforms really serve their purpose. I have worked as a freelancer and employee in various companies (Germany, EU) and have also been involved in several LC/NC projects. All of them failed due to the complexity of the business processes and failure to implement the requirements of decision-makers. The hoped-for simplification of implementation through UI-driven tools, which was supposed to save on devs, ultimately led to developers operating the UI tools and the business finding it too complex. However, the lc/nc SaaS solutions represented a discontinuity in the development
1/2
Aljoscha Rittner (beandev)
in reply to Aljoscha Rittner (beandev) • • •pipeline and were also considerably more expensive than traditional implementations, for which the developers had been trained.
So if lc/nc doesn't really work, I don't know how LLM with agent-based AI will help us. Plus, it's not even deterministic. The lc/nc platforms even have the advantage of working with code in the background that was created by developers who hopefully still understand it. Unfortunately, the same cannot be said for many prompt hunters who generate code.
I am very skeptical about what the future will bring with AI. But this is my personal opinion and experience (and I'm not alone in my area). However, as a professional I use AI to do boring stuff faster. Like a wrench or hammer. But not as a machine to produce complete solutions.
2/2
@evan
Jaylbird
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I didn't acknowledge in this discussion that the special role that software developers and other technologists have doesn't just mean extra privileges and higher wages. For some people, it's a lifeline.
Some neuro-atypical, LGBTQ+, politically dissident, disabled, culturally non-conforming and simply weird and eccentric people have found a place in society where they are tolerated and even celebrated because of their ability to make and maintain software systems.
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I don't think it's ever been a safe harbour, but it's been a handhold.
I think it's more than reasonable for people to question what the growth of AI-assisted and AI-generated software development and maintenance means for that handhold.
Marcus "MajorLinux" Summers
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •By far, one of the best takes Iβve seen on here.
I fear that the tech world has become too cliquey.
I appreciate the βfound familyβ aspect of what the space has become, but it turns to tribalism & fandom real quick because a lot of people have internalized the βus vs themβ mentality as a means of safety that has turned quite toxic.
The humanity of others has been lost on the internet for so long and a lot of the hurt is starting to come from people who have been hurt the most.
Rick the Card
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Have you heard of this? I'm curious about it.
makululinux.com/
Makulu World β Linux AI OS & Electra AI Center
Makulu WorldEvan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Rick the Card • • •Rick the Card
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Thomas Lee β
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Thomas Lee β • • •Nicolas Constant
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •common GenAI images aren't great, since they're generated by image uneducated users.
But I've seen really interesting work by photographers already (it's not a surprise, since their own art got the same criticisms a century ago "it's not your work, it's the machine's", "it's not art, it's engineering"). But that also implies that a lot of curation and post-production takes place in the process, just like photography.
And of course, using the medium to say something through it.
Thomas Lee β
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Thomas Lee β • • •Thomas Lee β
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •ΠΠ΅Π½ Π‘Ρ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •David Penfold
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to David Penfold • • •Dr. Ferrous
in reply to David Penfold • • •David Penfold
in reply to Dr. Ferrous • • •@DrFerrous
It's a glorified parlour game, and I think it will generally be seen as such soon enough.
Caveat: smaller, more precise, models may end up replacing LLMs for specific tasks and I think they'll be a lot better.
Rich Stein (he/him)
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Sieva π΄πποΈπΉ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Data doesn't support your belief.
βglobal electricity consumption for data centres is projected to double ... by 2030 ... representing just under 3% of total global electricity consumptionβ
βmainly driven by AI adoptionβ
iea.org/reports/energy-and-ai/β¦
I'm sure you can find more online. Or even ask you agent of preference π Your example simply ignores the big picture, and narrows down the problem to someone's personal choice.
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Sieva π΄πποΈπΉ • • •my previous job was building greenhouse gas inventories at the national, state and city levels.
Electricity generation is about 1/3 of global emissions; 3% of that would be less than 1% of global emissions for all data centres.
Even given the numbers you're sharing, and even if AI were responsible for all data centre energy use, AI is not a major source of GHG emissions compared to transportation (15%), cattle (7%), cement production (8%), deforestation (15%) or rice cultivation (3%).
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •@Anibyl there are a lot of great sites that give a good accounting of global emissions. EDGAR has a pretty accessible site:
edgar.jrc.ec.europa.eu/climateβ¦
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Sieva π΄πποΈπΉ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Sieva π΄πποΈπΉ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Sieva π΄πποΈπΉ • • •@Anibyl it's absolutely whataboutism.
I started the conversation talking about relative contributions to global GHG emissions. That requires comparing different activities.
If you want to help stop climate change, skipping one hamburger does 2000 times more good than avoiding AI.
Meg Ford
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •βThe rush to build supply has bid up the cost of conventional gas-fired plants (and deliveries are back-ordered for 5+ years). To work around that, big tech companies are bringing small truck-mounted turbines designed for short-term or peak use by the dozens. This is the opposite of economies of scale. These generators have roughly 50% higher fuel costs than a utility-grade combined cycle gas-burning plant, and the maintenance of a field of engines is multiples more expensive. That also means 50% more emissions pushed onto the local community, and tempting opportunities for developers to play fast and loose with air permitting rules.β
blog.ucs.org/mike-jacobs/as-seβ¦
As Seen in State of the UnionβUtilities Bend Under Too Much Demand
Mike Jacobs (The Equation)Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Meg Ford • • •Meg Ford
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I think the question is not so much about current pollution but what will happen as things scale. βGiven the direction AI is headedβmore personalized, able to reason and solve complex problems on our behalf, and everywhere we lookβitβs likely that our AI footprint today is the smallest it will ever be. According to new projections published by Lawrence Berkeley National Laboratory in December, by 2028 more than half of the electricity going to data centers will be used for AI. At that point, AI alone could consume as much electricity annually as 22% of all US households.β
technologyreview.com/2025/05/2β¦
ai-energy-usage-climate-footprint-big-tech
www.technologyreview.comEvan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Meg Ford • • •aeva
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Scanner
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •GitHub - scanner/booklore: A local-first RAG (Retrieval-Augmented Generation) platform for querying fiction books.
GitHubEvan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Scanner • • •Scanner
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •kini
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •julian
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
Unknown parent • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
Unknown parent • • •Mark Andrew π¨π
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •yeah. It's good for bouncing design ideas off of. But for producing code it's basically automated stack exchange.
Which explains why Jeff Atwood hates AI.
Aslan French
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Aslan French • • •@jcklpe.bsky.social I think there are a lot of reasons for this.
One big reason is gatekeeping. For those of us who've made our living from building software, it's threatening to see people without our skills making software for themselves. Especially if we are emotionally and intellectually invested in the craft of software making.
JimmyChezPants π¨π¦
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Mike Blake π΅πΈπ±π§πΊπ¦β§οΈπ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Mike Blake π΅πΈπ±π§πΊπ¦β§οΈπ • • •lina isidora
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •my employer will pay for pretty much any tools one can think of and yeah. for a quick "hey should this fail and panic!() and crash everything according to the docs" llms are (mostly) great
not so great when the thing i'm trying to do is niche and/or specific to the [in-house] libraries we use, but considering the main offender is mostly a thin api wrapper around an opensource c thing it manages more or less fine
would i want the whole "ai" industry to go up in flames without any hope for recovery? yeah
am i gonna take everything my BigTechCorpInc. employer will give me? also yeah
tho now that i think about it, maybe i could ask them for a 128gb ram mac mini or smth as a one-time purchase as an ollama server instead of the subscriptions
robcornelius
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •NIGHTEΝΆ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to NIGHTEΝΆ • • •NIGHTEΝΆ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •I see, thatβs totally fair!
And thank you for sharing your stance on the subject in general, itβs a very heated topic but I appreciate reading your balanced argument
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
Unknown parent • • •@jeremiah A "derivative work" is a term of art in copyright law.
en.wikipedia.org/wiki/Derivatiβ¦
Derivative work - Wikipedia
Contributors to Wikimedia projects (Wikimedia Foundation, Inc.)Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
Unknown parent • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
Unknown parent • • •Jonathan Lamothe
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • •@Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ This is where I'm going to have to respectfully disagree.
If I'm going to ship code, I need to be able to be able to answer for what that code does. The necessary level of understanding can be achieved with LLM-generated code, but it involves studying its output to be sure that every lineβevery statementβis understood. To me, it seems that the amount of work necessary to achieve that level of understanding negates any boost in prodictivity I may get from using the LLM in the first place, and it certainly doesn't justify the ethical cost.
I've been working on a fork of a large FOSS project lately. For me, the bottleneck hasn't been the actual writing of the code. More than half of my time has been spent studying the existing code to understand how it works so I can figure out the best way to make my edits. An LLM can't just magically give me that understanding.
Also, when it comes the actual writing of code, it's easier for me to express what I want in the unambiguous syntax of a programming language rather than the crude tool that is natural language. The latter I only use for communicating with other humans, and only because I lack a better tool for doing so. I'll freely admit that that last point is not universally true for everyone.
Also, I'll concede that the use of third-party libraries muddies the waters in terms of understanding how the code works. This is why I choose not to crucify those who do use LLMs to write code. I just choose not to participate.
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Jonathan Lamothe • • •@me Hey, Jonathan. You are not disagreeing with me.
You should go back and re-read what I said about how I use LLMs. I don't "vibe code", and I don't let LLMs generate a lot of code for me. I use LLMs for reference, rubber-ducking, and occasionally to do mechanical transformations (like changing the order of arguments to a method).
Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Jonathan Lamothe
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • •@Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ I guess my problem here is that I can't trust the LLM not to "hallucinate" the answer. I would feel the need to double check them, negating any benefit I might get because I'd have to do the same amount of work anyway. I can perhaps see some limited utility in using the LLM as a sort of search though, e.g.: "where in this codebase can I find the logic that handles $thing?"
The one point where I really get hung up is this: if machines can in fact thinkβwhich I doubt for reasons that are outside the scope of this postβwould that not make it morally wrong for us to try to bend them to our will? What makes that different from slavery? I bring this up because it seems to be the argument supporting your position that LLM output doesn't constitute a derived work of its training data.
Jonathan Lamothe
in reply to Jonathan Lamothe • •Jonathan Lamothe
in reply to Jonathan Lamothe • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Jonathan Lamothe • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • • •@me I don't think that LLMs have to be persons for LLM output to not be derived works of the training data! Also, it's not my place to decide, but cases like Doe vs. GitHub seem to be leaning that way.
en.wikipedia.org/wiki/Doe_v._Gβ¦
Doe v. GitHub, Microsoft, and OpenAI - Wikipedia
Contributors to Wikimedia projects (Wikimedia Foundation, Inc.)Jonathan Lamothe
in reply to Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ • •@Evan Prodromou π¨π¦πΊπΈπ¬π·π΅πΈ I'll have to have a look at that, though: