While I understand where this author is coming from. and agree that we’ve had a number of false starts or silly concepts like “prompt engineer”, there is definitely a skill in using LLMs, and it is a spectrum for different people.
Being completely engrossed in the new trends is probably not necessary, but keeping abreast of what’s possible and trying new things is. People who have an ignored/avoided LLMs completely are not able to just pick them up today and be as productive as people who have been learning the best ways to use them over the last couple years.
Many of the “skills” are not new, it’s the same skills you’ve been using throughout your career, just with a new tool. But any new tool requires practice and learning what it’s capable of.
I’m constantly experimenting around the edges of what LLMs are good at because it keeps changing and just because an LLM wasn’t good at X task 6 months ago doesn’t mean it still isn’t good at it today. The only way you can know is by testing and trying out things.
I work with a number of smart people, but their LLM proficiency is a wide spectrum. LLMs might/are a leveling factor, but there is still skill in getting what you want out of an LLM. This is most apparent to me when I interact with people who aren’t developers or aren’t as technically savvy. They just aren’t able to produce or get the same things out of an LLM that a developer or more technical person can.
> People who have an ignored/avoided LLMs completely are not able to just pick them up today and be as productive as people who have been learning the best ways to use them over the last couple years.
What exactly are you challenging? All he's saying is that people who have experience using a complex and changing technology are more effective with it than people that don't have experience. This seems like an uncontroversial take.
In my normal life, most people around me are not using AI and don't really know what to make of it. I tend to ask three questions:
* Do you think AI could be useful in your career?
* Do you think leveraging AI to achieve good results requires skill?
* How long do you think it would take you to develop those skills?
Everybody I've had this conversation with has answered yes to one and two, and everyone seems to think it'll take them between one and six months to develop those skills. I've been using AI for about two and a half years on a daily basis, and I'm still learning things daily, so I think that's the part we're probably underestimating.
Tower? This is from first-hand experience. I see it within my company. I’m not saying you can’t learn or won’t “catch up”, I’m simply pointing out that using LLMs effectively is a skill you need to hone.
- Learning in a self-directed way that aligns with the needs of your role and where it is going
- Clear communication, the ability to edit and know how to say what is important
- Forming and testing hypotheses
- Evaluating and testing tools and work product, and being able to recognize and rectify deviations
- Delegating, setting constraints, rubrics and escalation criteria
- Having a clear definition of done and being able to stick with it
- Understanding business domains and modelling them usefully
- Collaborating with others, communicating your needs and capabilities clearly, eliciting the needs and capabilities of others, synthesizing conflicting needs and aligning compounding capabilities
The adoption of genAI and agentic tooling really just serves to make all of these more important.
I basically agree with you, even though I think AI is a bad thing. But I think your list should be qualified. AI itself modifies expectations and changes the nature of the game, so that people care less about rigor in the way you described. With AI, since it's so easy to make new apps, people will care less about whether an app is correct or well-defined. And over time our expectations will change and our reactions will be gamed so that the person who makes the app the quickest with the most flash will win over those who put more care into things.
There's always been a tension between rigour and marketing of course but AI is a new order of magnitude shift towards the marketing and hype side of things. So I think while your skills are still needed, they will be overshadowed by a new set of skills to manipulate the audience or userbase to use the suboptimal more than was possible before.
You might be assuming that the skill is standalone and that the market will see a goldrush that drives the price to near zero.
An analyst could learn enough SQL in a few weeks to make their workflow better (leaving aside that AI analyst is already a thing).
Almost anyone could spend a few weeks getting a heavy truck license. Barriers to entering this trade are not that high but earnings have stabilised above subsistence level.
I'm sure there are plenty of analysts already using chatbots to improve their workflow. And the barrier to entry to heavy trucking isn't the license, but being being willing to not only drive a massive friggin truck, but to do so for many many hours a day often entailing being away from your family/relations for the vast majority of time. It's a tough life, and that's the reason the stereotype of truckers is what it is.
By contrast the LLM 'skill' of 'sit on ass and talk' is something basically everybody already does. So you just some minimum level of general cognitive ability and you're good to go.
This is not my experience at all. I see more people having difficulties in extracting value out of AI than the other way around. When I check what they're doing, they paint themselves in all kinds of weird corners, like keeping a mega-session full of unrelated questions, letting the AI model lead instead of managing it, misunderstanding the scope of what to ask for, etc.
Of course all the crap about prompt engineering or whatever is meaningless, and it always was. That said, this idea that if AI gets good enough you will just jump in and use it is flawed. You learn by doing. It's almost self-evident to me that someone who has been working with AI models for the past 4 years will be much better at it than someone who started this week.
"Getting it right is more valuable than getting in first" is what I will agree with. I feel what the author is trying to say. The AI field is moving too fast. One day it is prompt engineering. Then it's RAG, MCPs, Loop engineering, and now harness engineering. But rather than fearing that these are becoming obsolete, I feel there is an opportunity to participate by hopping on and sprinting ahead.
On what basis are you telling this? I think, it's more "outdated" skills that are worthless, but the interval of time we have to refresh those is getting ridiculously fast....
1-2 years ago we were told even by people like simonw that if we do not get on the AI train ahead of time we will never be able to catch up with the AI experts who had years of experience with AI.
Meanwhile in practice the opposite happened. You just need bog standard software development experience with an agent focus. You need to know how to set up an agentic CI pipeline, how to set up on demand sandboxes, basically a lot of devops and software development life cycle knowledge and less AI specific knowledge. It's just that AI caused those old things to become ever more important and urgent.
So what the author means by "AI Skills" are skills that relate exclusively to AI and no other domain.
Wow, this is a great point. I guess there's two parts to this in my mind: the media part and the actual impact on the developer population. The media/tech influencer part is unsurprising, anyone publishing media in the AI space has to constantly push hot and FOMO inducing takes in order to stay relevant. The impact on devs is more interesting; the AI has become so accessibly good by default that everyone is expected to be able to speed run every part of the stack and infrastructure.
Which is by the way the reason why we will probably see some errors on even bigger scale than before. We will forget to check things, rush, push to prod to early. Sometimes those slow checks were key moments.
> But a skill that can be picked up in a few weeks, and will be obsolete in a few months, has no value outside the moment.
This does not describe the general skill of effective communication, which is a prerequisite to writing a prompt that actually describes what you have in mind. Nor does it describe the general skill of, well, having something coherent and sensible in mind.
If anything, we are getting worse at acquiring these skills, thanks to an overreliance on AI. Everything turns to slop, if we let it.
Correct, AI skills became worthless in 2026, because nobody gives a damn about how good your prompt is. Your boss and customers only care about how the output positively or negatively affects them and as far as we know, the standards for what constitutes good output have been dropping.
> a skill that can be picked up in a few weeks, and will be obsolete in a few months
What skills are being referred to?
What's an example of such a skill that "can be picked up in a few weeks"?
The only "AI" example they seem to give is:
> in five years, I’m sure there will be old heads doggedly adding 'you are an experienced software developer, make no mistakes'
1. I don't think "make no mistakes" was ever considered a real AI skill - it was popularized as meme to mock those who did not know much about AI?
2. In years things may change, but Googling seems to indicate that mentioning a role in a prompt can be helpful as it quickly sets the stage for the AI to understand the persona/purpose/context/scope/desired response format for the rest of the prompt. On the other hand, I read that Anthropic found personas didn't boost result quality. But I'm still not sure it fits the article's core claim of "a skill that can be picked up in a few weeks, and will be obsolete in a few months". (Separately, even if not beneficial, that's also still not the same as "harmful" especially considering the few tokens it uses.)
On the other hand, Anthropic and OAI both still have live docs using the "You are a __" format:
Anthropic and OAI both have prompt guidance, including even version-specific guidance. I guess that might closest fit the article's definition, but personally I haven't heard about anyone taking "weeks" to learn it as a "skill", or at the very least I think most people know (since it's version specific) that it's not long-term.
Personally I frequently use different open models so it makes less sense for me, but I do sometimes ask LLMs for prompt feedback, especially for items I reuse frequently over months or more.
The criticism of the "prompt engineer" craze years ago would be valid, but I don't think criticizing that is relevant in 2026.
While I understand where this author is coming from. and agree that we’ve had a number of false starts or silly concepts like “prompt engineer”, there is definitely a skill in using LLMs, and it is a spectrum for different people.
Being completely engrossed in the new trends is probably not necessary, but keeping abreast of what’s possible and trying new things is. People who have an ignored/avoided LLMs completely are not able to just pick them up today and be as productive as people who have been learning the best ways to use them over the last couple years.
Many of the “skills” are not new, it’s the same skills you’ve been using throughout your career, just with a new tool. But any new tool requires practice and learning what it’s capable of.
I’m constantly experimenting around the edges of what LLMs are good at because it keeps changing and just because an LLM wasn’t good at X task 6 months ago doesn’t mean it still isn’t good at it today. The only way you can know is by testing and trying out things.
I work with a number of smart people, but their LLM proficiency is a wide spectrum. LLMs might/are a leveling factor, but there is still skill in getting what you want out of an LLM. This is most apparent to me when I interact with people who aren’t developers or aren’t as technically savvy. They just aren’t able to produce or get the same things out of an LLM that a developer or more technical person can.
> People who have an ignored/avoided LLMs completely are not able to just pick them up today and be as productive as people who have been learning the best ways to use them over the last couple years.
How's it up there in that tower? Citation needed.
What exactly are you challenging? All he's saying is that people who have experience using a complex and changing technology are more effective with it than people that don't have experience. This seems like an uncontroversial take.
In my normal life, most people around me are not using AI and don't really know what to make of it. I tend to ask three questions:
Everybody I've had this conversation with has answered yes to one and two, and everyone seems to think it'll take them between one and six months to develop those skills. I've been using AI for about two and a half years on a daily basis, and I'm still learning things daily, so I think that's the part we're probably underestimating.Tower? This is from first-hand experience. I see it within my company. I’m not saying you can’t learn or won’t “catch up”, I’m simply pointing out that using LLMs effectively is a skill you need to hone.
"AI skills" are the same skills we always needed:
- Learning in a self-directed way that aligns with the needs of your role and where it is going
- Clear communication, the ability to edit and know how to say what is important
- Forming and testing hypotheses
- Evaluating and testing tools and work product, and being able to recognize and rectify deviations
- Delegating, setting constraints, rubrics and escalation criteria
- Having a clear definition of done and being able to stick with it
- Understanding business domains and modelling them usefully
- Collaborating with others, communicating your needs and capabilities clearly, eliciting the needs and capabilities of others, synthesizing conflicting needs and aligning compounding capabilities
The adoption of genAI and agentic tooling really just serves to make all of these more important.
I basically agree with you, even though I think AI is a bad thing. But I think your list should be qualified. AI itself modifies expectations and changes the nature of the game, so that people care less about rigor in the way you described. With AI, since it's so easy to make new apps, people will care less about whether an app is correct or well-defined. And over time our expectations will change and our reactions will be gamed so that the person who makes the app the quickest with the most flash will win over those who put more care into things.
There's always been a tension between rigour and marketing of course but AI is a new order of magnitude shift towards the marketing and hype side of things. So I think while your skills are still needed, they will be overshadowed by a new set of skills to manipulate the audience or userbase to use the suboptimal more than was possible before.
Every skill that any literate person can get in a few weeks is worthless. Always was the case.
You might be assuming that the skill is standalone and that the market will see a goldrush that drives the price to near zero.
An analyst could learn enough SQL in a few weeks to make their workflow better (leaving aside that AI analyst is already a thing).
Almost anyone could spend a few weeks getting a heavy truck license. Barriers to entering this trade are not that high but earnings have stabilised above subsistence level.
I'm sure there are plenty of analysts already using chatbots to improve their workflow. And the barrier to entry to heavy trucking isn't the license, but being being willing to not only drive a massive friggin truck, but to do so for many many hours a day often entailing being away from your family/relations for the vast majority of time. It's a tough life, and that's the reason the stereotype of truckers is what it is.
By contrast the LLM 'skill' of 'sit on ass and talk' is something basically everybody already does. So you just some minimum level of general cognitive ability and you're good to go.
This is not my experience at all. I see more people having difficulties in extracting value out of AI than the other way around. When I check what they're doing, they paint themselves in all kinds of weird corners, like keeping a mega-session full of unrelated questions, letting the AI model lead instead of managing it, misunderstanding the scope of what to ask for, etc.
Of course all the crap about prompt engineering or whatever is meaningless, and it always was. That said, this idea that if AI gets good enough you will just jump in and use it is flawed. You learn by doing. It's almost self-evident to me that someone who has been working with AI models for the past 4 years will be much better at it than someone who started this week.
"Getting it right is more valuable than getting in first" is what I will agree with. I feel what the author is trying to say. The AI field is moving too fast. One day it is prompt engineering. Then it's RAG, MCPs, Loop engineering, and now harness engineering. But rather than fearing that these are becoming obsolete, I feel there is an opportunity to participate by hopping on and sprinting ahead.
On what basis are you telling this? I think, it's more "outdated" skills that are worthless, but the interval of time we have to refresh those is getting ridiculously fast....
> On what basis are you telling this?
A career moat, probably. And I agree.
1-2 years ago we were told even by people like simonw that if we do not get on the AI train ahead of time we will never be able to catch up with the AI experts who had years of experience with AI.
Meanwhile in practice the opposite happened. You just need bog standard software development experience with an agent focus. You need to know how to set up an agentic CI pipeline, how to set up on demand sandboxes, basically a lot of devops and software development life cycle knowledge and less AI specific knowledge. It's just that AI caused those old things to become ever more important and urgent.
So what the author means by "AI Skills" are skills that relate exclusively to AI and no other domain.
Wow, this is a great point. I guess there's two parts to this in my mind: the media part and the actual impact on the developer population. The media/tech influencer part is unsurprising, anyone publishing media in the AI space has to constantly push hot and FOMO inducing takes in order to stay relevant. The impact on devs is more interesting; the AI has become so accessibly good by default that everyone is expected to be able to speed run every part of the stack and infrastructure.
Which is by the way the reason why we will probably see some errors on even bigger scale than before. We will forget to check things, rush, push to prod to early. Sometimes those slow checks were key moments.
> But a skill that can be picked up in a few weeks, and will be obsolete in a few months, has no value outside the moment.
This does not describe the general skill of effective communication, which is a prerequisite to writing a prompt that actually describes what you have in mind. Nor does it describe the general skill of, well, having something coherent and sensible in mind.
If anything, we are getting worse at acquiring these skills, thanks to an overreliance on AI. Everything turns to slop, if we let it.
Correct, AI skills became worthless in 2026, because nobody gives a damn about how good your prompt is. Your boss and customers only care about how the output positively or negatively affects them and as far as we know, the standards for what constitutes good output have been dropping.
> a skill that can be picked up in a few weeks, and will be obsolete in a few months
What skills are being referred to?
What's an example of such a skill that "can be picked up in a few weeks"?
The only "AI" example they seem to give is:
> in five years, I’m sure there will be old heads doggedly adding 'you are an experienced software developer, make no mistakes'
1. I don't think "make no mistakes" was ever considered a real AI skill - it was popularized as meme to mock those who did not know much about AI?
2. In years things may change, but Googling seems to indicate that mentioning a role in a prompt can be helpful as it quickly sets the stage for the AI to understand the persona/purpose/context/scope/desired response format for the rest of the prompt. On the other hand, I read that Anthropic found personas didn't boost result quality. But I'm still not sure it fits the article's core claim of "a skill that can be picked up in a few weeks, and will be obsolete in a few months". (Separately, even if not beneficial, that's also still not the same as "harmful" especially considering the few tokens it uses.)
On the other hand, Anthropic and OAI both still have live docs using the "You are a __" format:
https://platform.claude.com/docs/en/build-with-claude/prompt...
https://developers.openai.com/api/docs/guides/prompt-enginee...
Anthropic and OAI both have prompt guidance, including even version-specific guidance. I guess that might closest fit the article's definition, but personally I haven't heard about anyone taking "weeks" to learn it as a "skill", or at the very least I think most people know (since it's version specific) that it's not long-term.
Personally I frequently use different open models so it makes less sense for me, but I do sometimes ask LLMs for prompt feedback, especially for items I reuse frequently over months or more.
The criticism of the "prompt engineer" craze years ago would be valid, but I don't think criticizing that is relevant in 2026.
https://platform.claude.com/docs/en/build-with-claude/prompt...
https://developers.openai.com/api/docs/guides/latest-model