How to Survive in the Tech industry in 2026
blog.phuaxueyong.com
blog.phuaxueyong.com
I am not intrinsically opposed AI. I am opposed to its environmental and social impacts. I constantly see spamming AI bots that send zillions of useless or poor-quality PRs to OSS. I see PRs that add features, but are really huge, making maintenance even harder in the future without the help of AI. AI creates bloatware everywhere. I see AI trained on stolen data without respecting licenses. I see data centers popping up at a scale never seen, consuming more and more energy. more and more resources (They basically consumed all RAM/SSD of 2026).
"It drives rising energy prices in poor communities, disrupts wildlife and fresh water supplies, increases pollution, and stresses global supply chains. It re-enforces the horrible, dangerous working conditions that miners in many African countries are enduring to supply rare metals like Cobalt for the billions of new chips that this boom demands. And at a moment when the climate demands immediate action to reduce our footprint on this planet, the AI boom is driving data centers to consume a full 1.5% of the world’s total energy production in order to eliminate jobs and replace them with a robot that lies." [0]
I am tired of people that have no concerns about climate change or the impact of their collective actions on other humans. I hope that one day everyone will be judged for their role in this system.
[0] https://drewdevault.com/2026/03/25/2026-03-25-Forking-vim.ht...
It's also very destructive in a societal and environmental level. I didn't pick this music though. The only thing I can do is either dance or stand around while it plays.
> I am tired of people that have no concerns about climate change or the impact of their collective actions on other humans. I hope that one day everyone will be judged for their role in this system.
I have a more practical approach - I reject that individual action is meaningful to fight climate change.
Ir requires societal and political action for anything meaningful that might move this needle. Individual action will do nothing to fix systemic problems, you will only giving yourself a debuff for nothing.
Vote for parties that want to tackle the issue seriously, support initiatives at societal and political level that may so something in the right direction, etc.
"I am not using AI because it has a negative effect in climate change" will do nothing while a gargantuan amount of money is pumped into it so GPUs may turn electricity into bullshit text, code of questionable quality, and images of dubious taste.
It also seems to me that some people are leaning on them for things they should not be. It can help you research psychology; it cannot be your therapist.
In these strange times, apparently that makes me “resisting” or “a Luddite.”
proceeds to discount parent comment with handwavey argumentation
Is there a possibility of using automation to fix this very issue? Before you admonish me of burning more carbon to address output of burnt carbon, hear me out.
For example: if these PRs are on GitHub, then we have a set of actions that:
- run a regression test suite to ensure there are no net new bugs.
- Then it verifies whether the PR has high quality tests for the feature they're implementing
- runs a suite of complexity, readability, security tests and gathers metrics about them, automatically closing poor quality PRs and even blocking accounts that submit them?
I argue these automation are great to have for any project - those that have AI contributions or not.
> I see data centers popping up at a scale never seen, consuming more and more energy. more and more resources (They basically consumed all RAM/SSD of 2026).
> consume a full 1.5% of the world’s total energy production in order to eliminate jobs and replace them with a robot that lies
From my perspective, this is an extremely reductive view of AI. My view is that we are fundamentally undergoing a shift in knowledge work and these are all costs that will bear fruit for decades to come.
Imagine freeways being built on barren land. In decades to come it will bear fruit entire new cities and farms. Right now all that construction for those freeways in progress alarm anyone thinking of the climate but the farms that will be possible thanks to these freeways will offset the debt and even support the cities around them.
Any talk of new AI datacenters is peppered with its humongous energy needs. A cursory internet search can point to that: https://www.iea.org/reports/energy-and-ai/energy-supply-for-...
> Global electricity generation to supply data centres is projected to grow from 460 TWh in 2024 to over 1 000 TWh in 2030
Just to be clear, 540TWh is the yearly energy consumption of developed industrialized countries. It would sit in between Germany and Canada as of today: https://en.wikipedia.org/wiki/List_of_countries_by_electrici...
I offered data. You offered vibes.
"Math ain't mathing". The nerve of these pricks.
The Wikipedia link I posted is not "me making shit up". That page is sourced from https://ember-energy.org/data/yearly-electricity-data/
You, on the other hand, provided no sources. You claim to be quoting directly measured usage. But, unlike me, you provided no sources.
It is clear who is the one making shit up here.
Just one more thing you were wrong about. Not that it matters, you being wrong, seemingly, is something that occurs very frequently.
https://i.imgur.com/qv75o9j.png
So where's your 12KW solar array?
Your Grok prompt for lewd anime waifus most certainly don't.
One major problem I see with the use of AI is that it will prevent people from building an understanding of <insert problem domain X here>. This will reduce people's ability to drive AI correctly, creating a circular problem.
I don’t really think this is a problem. AI is a tool, you still learn while using it. If you actually read, debug and maintain the produced code, which I consider a must for complex production systems, it’s not really that different compared to reading documentation and using Stack Overflow (i.e., coding the way it was done 10 years ago). It’s just much more efficient and it makes problems easier to miss. Standard practices of AI assisted development are slowly forming and I expect them to improve over time.
I’m now working in the backend, and SQL is a major requirement. Writing what I would call “normal” queries. I’ve been reaching for AI to handle this, pretty much the whole time - because it’s faster than I am.
I am picking up tidbits along the way. So I am learning, but there’s a huge caveat. I notice I’m learning extremely slowly. I can now write a “simple” complexity query by hand with no assistance, and grabbing small chunks of data is getting easier for me.
I am “reading, debugging, and maintaining” the queries, but LLMS bring the effort on that task down to pretty much 0.
I guarantee if I spent even 1 week just taking an actual SQL class and just… doing the learning, I would be MUCH further along, and wouldn’t need the AI at all. It’s now my “query tool”. Yeah, it’s faster than I am, but I’m reliant on it at this point. I will SLOWLY improve, but I’ll still continue to just use AI for it.
All that to say, I don’t know where the future goes - our company doesn’t have time to slow down for me to learn SQL, and the tool does a fine job - it’s been 1.5 years and the world hasn’t ended, I can READ queries rather quickly - but writing them is outsourced to the model.
In the past, if a query was written on stack overflow, I would have to modify it (sometimes significantly) to achieve my goal, so maybe the learning was “baked in” to the translation process.
Now, the LLM gives me exactly what I need, no extra “reinforcement” work done on my end.
I do think these tools can be used for learning, but that effort needs to be dedicated. In many cases I’m sure other juniors are in a similar position. I have a higher output, but I’m not quickly increasing my understanding. There’s no incentive for me to slow down, and my manager would scoff at the idea, really. It’s a tough spot to be in.
Also makes me think of lot of incidental learning that can go on. Like when looking at API docs noticing the other things. Might not be useful now, but could very well be later.
> AI is a tool
A more helpful analogy is "AI is outsourced labor". You can review all code from overseas teams as well, but if you start to think of them as a tool you've made too big promotions into management.
Bravo! IMHO, AI just underscores core high quality engineering practices that any high quality engineer has been practicing already.
AI is a tool that provides high leverage - if you've been following practices that allow sloppy coding, AI will absolutely amplify it.
If anything, I would guess that the AI assisted future will require engineers to think through the problem more upfront and consider edge cases instead of jump and type out the first thing that comes to mind - the AI can spit out code way faster.
There's an alternate, vibe coded universe where engineers just spit out slop but as I wrote in another comment here, there are tools to detect that. These are tools that sound "enterprisey" and that's because before AI, no one else had to deal with such scale of code - it's was just far too expensive to read, update and create PRs.
Those boundaries are coming down and now almost everyone who can pay for Oxygen tanks have a shot at scaling Mt. Everest.
Very much the opposite. LLMs do a fantastic job of increasing accessibility of knowledge.
They have wide exposure to content and is incredibly good at spotting patterns and suggesting both well established norms from the current domain and serendipitous cross domain concepts.
I feel the concern you share is LLMs expose new frontiers to people who otherwise might not even have imagined that frontier exists and then those people do a lazy or superficial job of it because they lack any internal motivation to do a deep dive on it
The amount of progress humans have made over 2000 years, especially the last 100, is just phenomenal. If anything, there's very little evidence that any X prevents people from Y. X certainly might not require people to Y anymore but I don't question what a motivated person can do just like I don't question what an unmotivated person can refuse to do.
It would help if you expanded on "it will prevent people from building an understanding of <insert problem domain X here>", but could one interpretation of it be "AI will prevent people from building an understanding of good software design patterns" because AI allows people to just vibe code and pay no attention to design patterns at all?
Where does the majority of an individual's general knowledge come from?
Where does the majority of an individual's _understanding_ come from?
Reading is part of your environment - you didn't read your way into a vacuum - the very tools you use to read right now (language, symbols, syntax) are environmental gifts. Something, even if you're consciously unaware of it, made you read those material. The specific script you read is part of your environment (it's unlikely you will know what symbols in Hindi mean if you didn't grow up in that environment or weren't biased towards it). Your physical brain structure depends on the languages you speak and will physically morph as you change those languages.
If you have absolutely no clue what "mutton paratha" is, you won't ever start reading about "mutton paratha" right away. (there's another thread we can discuss about human biology, original thought and serendipity but that is better done over email than in HN comments). If you can hear and see, a lot of your general knowledge comes from hearing and seeing too, even if you're consciously unaware of it or don't acknowledge it. If you saw your dad working on cars all your young life, you might not think you have any idea about cars but you might be surprised.
> Where does the majority of an individual's _understanding_ come from?
Understanding is more complex than knowledge. Understanding and knowledge are at different levels of abstraction but you cannot understand something you have no knowledge of. Thus knowledge is a pre-requisite to understanding but insufficient. With sufficient understanding, you can synthesize knowledge. Nikola Tesla synthesized knowledge about AC Induction motors - a knowledge that's now taught even in high schools from his understanding of EMF. However given the same knowledge of EMF, many people at the times of Tesla didn't have sufficient understanding or motivation or capacity to birth AC Induction motors.
You can be extremely knowledgeable but understand absolutely nothing. LLM's are, I argue, excellent examples of this - they have tremendous knowledge but depends on human intervention and discourse to tease it out (as great as SOTA LLMs are, they are not going to randomly design apps without an initial prompt which provides them agency and biases them towards autonomy).
(Again, I'm discarding some details that we can discuss over email. For example: I can strongly argue that LLMs have some understanding of what its knowledge is - like the semantic distance between "mutton" and "paratha" and how they can change across various manifolds in latent space)
If we agree on this take on understanding, then I argue that the majority of an individual's understanding comes from their consciousness, when they have exercised those neural networks and processed their knowledge to arrive at inferences and re-enforced (belief) or disassociation (misbelief) of assertions
This thread's getting a bit long, so I'm happy to continue this discussion via email, if you like. My email is ben@overmyer.net.
The second reason why I take issue with that line is, as I've said, the fact that few employers allow employees time "on the clock" to do anything at all that's away from screens, and saying "do it in your spare time, then" is adding insult to injury. I have a rich social life, hobbies, and am raising a family. What I'm observing is that this is not helping my career one bit, and that's perfectly fine. Not everything in life needs to be in service of one's career. But this is also the reason, why I do not have time, off the clock, to attend meetups and events.
The third thing I would notice is that it helps your career (again, speaking from a "code monkey" perspective here), less than you'd think. What is going to come out of the chance encounters at meetups and events? Maybe someone wants to hire you. Maybe someone wants to work with you more informally. If you sell your time in 40-hour-per-week blocks, none of this is a business opportunity you can capitalize on. If you're on a job, you've already sold your 40-hour-block, and have nothing left to transact with. If you're off a job, you need a new one, and you need it now, so you need to be more transactionally-minded than just investing time into chance encounters.
Now, there is a separate consideration that may enter into career planning, namely that one might try to evade the 40-hour-per-week payrolled-employee trap, and try to prioritize maximizing hourly rate over yearly compensation and do freelancing. But this sort of consideration, in my mind, is not properly the domain of career advice. Career advice is: "Here is some mistakes you should avoid. Avoiding them is always an option, no matter what your circumstances are, and by avoiding them, you will always have better outcomes." This is not that: It is simply not the case, that everyone can and should be a freelancer.
Are you in SF? If not, it would be hard to explain how much of an impact geography can have on your success in life.
> The third thing I would notice is that it helps your career (again, speaking from a "code monkey" perspective here), less than you'd think. What is going to come out of the chance encounters at meetups and events?
I'm unsure if you're being facetious - billionaires have been made of people who happened to be at the right place at the right time and you don't get there by staying at home. I mention the billionaires because you can look these up - I'm pretty sure there's a far larger volume of people who made far less.
If you discount the value of chance encounters, you've not yet had the opportunity to realize how random success is. You increase your chances by increasing your chances at random success. This is all probability theory and provable mathematically.
Moving to SF is only an option for the rich and privileged. Saying no to a solid paycheck that comes with a 40-hour workweek to make space for randomness is for the rich and privileged. Some of us are born rich and privileged, some are not. Some of us are born as extroverts, some as introverts. For some of us, putting off-hours to use for doing more work-related stuff ends up working out, for others it wreaks havoc on our ability to have hobbies, social lives, and families and is a surefire way to destroy happiness (and might still not help our careers).
"Everyone needs to move to SF and start prioritizing hustling over staring at their editors and compilers" is terrible advice. For a sizeable proportion of people it's not an option. For a sizeable proportion of people it's a surefire way to destroy their lives.
With the current surge, everyone is (expected to act as) a senior/mentor to a swarm of workers that lack interpersonal/business context.
It’s not a huge shift if you’re already deeply invested in business lore, but it’s unfortunately a brutal speedrun of skills that were previous slowly accumulated for new/junior hires.