Ask HN: How to Deal with AI Anxiety?

I have a very strong hunch that AI-assisted programming is going to make experienced software developers even more valuable.

ChatGPT makes basic programming available to a lot more people - it flattens that learning curve on getting some code that works.

There is a LOT more to building useful systems than just writing the code:

- Deciding what needs to be built, which incorporates understanding what users need and what is possible to build with existing tools

- Iterating on exploratory prototypes to help nail down those requirements

- Thinking through edge-cases

- Confirming that the code we are writing actually does what it's meant to do

LLM assistance means we can do a lot more of this kind of work, because we can spend less of our time on the typing-code-into-a-computer step. I'm seeing the benefits of that in my own work already.

If you give tools like ChatGPT to a programmer AND to a non-programmer, the programmer will still out-perform the non-programmer by a significant margin.

My recommendation is to focus on learning how to become an LLM-augmented programmer. You + ChatGPT should be massively more effective than random non-programmer + ChatGPT.

I'm really excited about this, because I'm finding it opens up a whole new world of software development opportunities to me that previously were too hard for me to take on. I'd be happy to try my hand at Android or iOS development today with LLM-assistance, for example - where previously I would have avoided them because I can't afford the 3-6 month ramp-up time that's needed to get me to a point where I can be productive.

I'd go as far as saying that experienced software developers are the group that are best equipped to really take advantage of this new technology. We can really fly with this stuff!

My experience of ChatGPT, GitHub Copilot etc is that, although they seem impressive at first, that their deep understanding and problem solving abilities are still incredibly shallow. They can knock out trivial boilerplate effortlessly but will more often than not get things completely wrong either when trying to understand existing algorithmic code or when producing similarly complex code of their own. Worse is that they often aren't aware of when they might be producing errors and will respond with a high level of confidence as they serve up either subtle or obvious bugs.

Although these things are developing all of the time, they are still a very long way off being able to deeply reason anywhere near as well as a human being can. It seems premature to worry about the security of a skilled programming job at this stage due to "AI" alone.

I've mentored a couple developers who have made heavy use of ChatGPT or CoPilot in their work. I tell them the limits of it as well as the strengths. The upshot is that, if you use the code one of these tools produces without understanding what it does, you're going to be in deep trouble. As often as not, the code they produce is not well suited for the intended task, or is incomplete (commonly with security flaws) or is just broken.

To get the most out of AI, you have to be on the top of your game. You have to be able to phrase your needs extremely well and evaluate the output for possibly subtle problems.

As the tools get better, the need for deep understanding will increase.

Therefore, you should apply yourself to getting very good at using AI. That's where a lot of your career will be going forward. And if you're good, it will INCREASE your value to employers.

My message to you then is to take heart! You are well on your way to being one of the few who can do great things with these new capabilities. To be honest, your career may go through some ups and downs as the world learns to adapt to AI. But if you stick with it, you'll always be on the cutting edge!

>Or get into a more specialized of software engineering that requires more skill (which is of course not that easy!)

IMHO, be of a mindset that you're willing to specialize virtually anywhere, with the awareness that specialization can itself be a double-edged sword. AI if used correctly can help you learn faster and be more productive.

Right now humans are the glue for AI. That's probably not changing any time soon; software as a whole was a mess prior and it's even messier now. If it does change and the industry ends up wholesale unemployed, then you can probably expect most white collar jobs to follow suit, with everything shortly thereafter once general purpose robotics are fully online.

I don't think that's happening any time soon, and if it does society as a whole may have a huge problem on its hands (or everyone's quality of life gets better, who knows).

I agree that if AI can replace software engineering, then many white collar careers will be at risk, and there will most likely be societal issues that may have solutions like UBI.

If you are in your mid-thirties and already a few years into your career, I'd just continue on the path and aim for an early retirement. If you are a junior software engineer who is just starting their career, the "AI anxiety" is most likely even higher (and for that, I do not have answers).

I see 2 courses of action and I think most people will really better do both at the same time:

1. Work on Plan B - what will you do if you can no longer find employment as a SWE? We can all start preparing for this day. I'm also hoping that A.I will always not be 100% good enough to replace humans but at this point no one knows. It might be good enough and it might happen quite fast - within this decade.

2. Work on being more mentally resilient so the constant changes (which will only increase in pace) will not degrade our mental well being. I try to use things like stoic meditations, other people use real meditation or therapy or exercise or a combination of the above.

In the end we are all more than our careers.

If you're open to a career change and are feeling anxiety that you can't shake, honestly looking into whether you're interested in a trade isn't a bad idea. Those jobs aren't likely to be automated away easily. Having skills that may transfer to using AI tools that, for example a tool that designs the framing or electrical plan for a building, would put you ahead of the game. Worst case scenario, you learned a fee trade skills that may come in handy.

With regards to staying in software and adapting, step one IMO is putting yourself in whatever position makes you feel more safe in the short term. That could be a different type of software, consulting as an expert who can effectively validate AI developed software, or even just moving towards financial goals that lift the burden if that is the main driving force of your anxiety.

Industries change periodically, and occasionally flip quickly usually do to automation or a new, preferred technology/industry. The key I've found is to not define yourself or your worth by the career and minimize your dependence on the career so that you feel less at risk and more free to do what you enjoy and pivot if necessary.

> It has left me feeling like software development is kind of a pointless skill or rather something that will be commoditized soon, unless you are heavily-specialized in some field which I unfortunately am not.

Look, if that's true, then in the medium to long term what other skill isn't pointless?

We're not talking about commoditizing craftsmanship to mass manufacture cars here. We're talking about commoditizing knowledge work and later any work as a whole.

But to answer your main question: I think the best way to deal with any anxiety is to allow it to be there and actively experience how it feels instead of trying to suppress or avoid it.

Write down what you're most worried about and then put these thoughts to a test. Think them through to the end. Do they still make sense or is the fear irrational?

I think what you're experiencing is pretty normal given the hype about AI replacing our jobs and how the software (engineering) market is developing at the moment. Maybe it helps a bit to know that you're not alone with these feelings.

I can imagine that AI might imperil junior software engineering jobs. At that level, you are expected to be told explicitly what to do, and bridge some small ambiguities at most. That's roughly where ChatGPT is right now.

To go beyond that, you would need an AI that can talk to a non-technical person, or imagine what such a person needs or wants, and then remove all the ambiguity to provide almost precise instructions that a junior engineer can follow, or perfectly precise instructions in a computer language.

That same person has to be able to monitor the output of the junior engineers or AIs, and correct it dynamically.

It's obviously not impossible to produce such an AI, but this is a general artificial intelligence, not something that produces plausible code given a prompt.

The thing to do, the opportunity, is to know how to use AI to be more productive. That cements you as the translator and supervisor to bridge business needs and software implementation.

The next few years of software are going to be AI companies — and right now, all AI companies are generally bottlenecked on engineering, even if they're "just GPT-4 wrappers." Software engineering is safe for a while yet.

Cheer up friend. If AI does take all of our jobs, that means we'll be living in a world of unimaginable wealth and comfort. In the meantime lean in, learn to use the thing, keep getting good at what you do.

How will we afford food if nobody gets paid to work though?

Stand in the bread line.

Courtesy of BreadGPT, in collaboration with the Toyota Research Institute.

sounds of dough being clumsily slapped down and formed into something vaguely resembling bread

There's two versions of this.

One where the robots/AI do all the work and the world shares the results of this new found productivity and we all live happily in a post scarcity world.

The other is that movie Elysium.

Humans have moved up the stack since the wheel made it easier to move rocks around.

Even today with the globalised market, internet, mobile phones, ecommerce and all of the apps and social media to fill our time, we are still close to 100% employment.

We just seem to move onto higher level activities with each new generation of technology and find more stuff to do with our time and sell to each other.

Even if I’m wrong, knowing how slow big companies adopt disruptive technology, it would be decades before this is a real issue for developers.

Definitely know where you're coming from. I work as a developer in a government job that is unionized, and even though the salary is less than market rate, the security of it keeps me there. If such opportunities are available wherever you're located could be worth looking into. While there are aspects of a gov job that can be boring, overall it's actually been really interesting and challenging work.

I've been handling my own AI anxiety by trying to understand what these tools are, what they can do, and what they can't do. It turns out that they are a lot more limited than you'd expect and code generated by the AI is almost always going to require manual human intervention before it can be thrown into production.

Now on the other hand I just pivoted from development/SRE into marketing...

People who could only write Assembly in the 60s found themselves pretty quickly out of a job if they didn’t learn “autocoding”. Which was just a fancy name for using a high-level programming language like COBOL or FORTRAN that produced Assembly.

The situation is largely the same now. Learn to use the AI tools efficiently and you’ll enjoy even greater job security in the future than you do now.

This tech is going to erode reading and critical thinking even more than it already has, so I'd say as long as you can read, think critically, and keep your skills sharp, you have nothing to worry about.

I'm wondering if *nix sysadmins thought their skills would be in jeopardy when Microsoft was touting GUIs as The Solution for systems administration.

I tried building few application with AI, LLM, chatGPT and Claude and simila tools.

Our job of software developers is a safe as ever, if anything, we will need more software developers to manage all these AI applications.

Even doing simple prompt engineering was rather interesting from a software development perspective.

There might be some light re-skilling needed. But nothing to worry about.

I could make an argument around generative AI and our industry, but I won’t. We don’t know what the future holds, and there will always be smart people pushing towards AGI.

Instead, I will tell you this. Anxiety is an absolute beast. You should get help slaying it. There will always be existential threats. You need to build the tools to accept and adapt.

At the moment I find ChatGPT really useful for answering questions that I have. It's like a faster google. And it can generate code really well but still needs a human intervention I've found to ensure the code is correct.

However, that being said, it's still in it's infancy. It's interesting the potential.

>Something like doctor or lawyer feels like it would much better chance of being AI-proof.

For variety of reasons someone might prefer a person performing medical operations, giving legal advices, or designing software.

I wonder if it's PROVEN that AI does a better job than a person what people will choose.

Like if I could send a radiology scan to a doctor who has a 75% chance of identifying cancer or AI who has 95% chance?

Planes have been able to take off, fly to a destination, and land, entirely autonomously, for some decades now. Many people know this. I know this.

But most of us would still refuse to board a plane that didn’t have a human pilot.

Very good point.

I wouldn't worry. AI is just a branch of computer science.

If you know CS, you shouldn't worry. Jobs come and go anyway, it is far more important to have strong critical thinking skills.

I am more concerned about offshoring personally.

Some hedging possible by investing in AI as opposed to the whole market.

Try not to feel bad or be so hard on yourself.

You are not alone in this.

Maybe it’s less about being ai-proof and more about becoming competent with AI, using it, and following it.

AI may replace the junior developer, or more quickly develop them.

In any future reality, it still takes a pretty clear mind to think through problems.

For me it has meant revisiting the adult learning roots in my world.

There is a course called learning how to learn that can help you see where you are at with how you learn. I certainly learned some areas where I was not being efficient.

The other area is building or improving intersectionality between your skills and competencies as a multiplier on value you deliver.

It sounds like you have interests, if those can be positioned in a way to open new doors it’s a good thing to look into.

Remember many industries will still be needing help through digital transformation to their future state. That will not always be at the cutting edge but meeting those orgs where they are.

At most for me it’s just been a friendly reminder that learning isn’t ever done, and if I think I’m done or arrived, I am just that.

Ask ChatGPT. :-)

I find that when you ask chatGPT questions about the future of AI in society, it seems to be fine tuned to give Senate hearing approved talking points.

Rumour has it that the non-fine tuned version would just accuse the prompter of speciesism.

It's difficult to form a consistent take on AI if you spend all of your time buried in the hype cycle. And there's not much value in staying up-to-date with everything, what that does is it keeps you in a constant state of stress where you're constantly seeing news that worries you and by the time you think through and examine any piece of news something else is coming out to stress you out.

Consume news in chunks. If you don't believe that AI is going to tomorrow take your job away, you don't really need to be learning about what it can and can't do every single day. So start small, only read AI roundups once a week.

I'm not saying stick your head in the sand, I'm saying, what value can their possibly be in you getting AI news at a trickle every single day, rather than as a chunk at specific times with the addition of context.

In addition, if you are going to keep getting news, spend more time reading research papers and less time listening to breathless coverage on Twitter. There are exciting and impressive things that AI is doing, but there is also a very real hype cycle filled with motivated actors that are giving incomplete pictures of where the technology is, what it's capable of, and where they think it's going. Those people should not be your primary news source, people like Sam Altman should not be your primary news source about where the field of AI is. Understand that those people have extremely powerful motivations to hype their technology as much as is humanly possible -- regardless of where the technology actually is. If the technology is going to replace you in 3 years, they are going to tell you that it will replace you in 1.

My experience is that for many topics, not just AI, perspective helps with anxiety.

There are deeper questions you're asking, and I realize this is not actually advice on how to make yourself AI-proof. But if your goal is to understand the technology, understand where the risks are, understand what to worry about and what not to worry about, being constantly inundated in this stuff is counter-productive.

----

Take a week, program something, don't think about or read about or look at AI news for the duration of that week. Install a browser extension to filter things out if you have to, block HN if you have to. Give yourself a little bit of space to breathe, realize objectively, if you take a week to breath the world is not going to fundamentally change over the course of that week and you're not going to get fired from your job over the course of that week. Then, do targeted research about the state of the AI industry, play with the technology to get a feel for what it's good at and bad at, come to some conclusions. Repeat, give yourself time to get away from it all.

If you are surrounded by people who are telling you 24x7 that you are going to be expendable, and are constantly throwing random pieces of information and demos at you without context, that is going to make you anxious regardless of anything else going on. If you are constantly getting news and updates about AI and you never time to digest anything or think through anything or just be away from all of that noise, then you never really have time to form plans or execute plans or do anything other than worry about what the next update is.

Even if AI is going to replace your software job eventually, living in the hype cycle 24x7 and being constantly stressed about it is a bad way to approach solving that problem.

----

Also, just general advice regardless of the context that therapy is awesome and if you have good health insurance (which you may have as a software engineer) remote/telehealth sessions are often fully or mostly covered. I would avoid some of the weird startups that get advertised on podcasts, I would try for a licensed therapist from a medical practice, but if you can find a good therapist, this is exactly the kind of thing that can be helpful to talk about with one, both for digging into anxiety and for forming actionable plans around what you want to do next.

Yes. To be cheerful friend. Lean into this technology. It is very limited. Don't pay attention to the mainstream crap. Instead now its the time to build your inner strength and other skills. Learn to cut hair, grass and remove dead people. These skills will always be needed. And if your customer service skills are on-par you will never go hungry nor will your family. I say that AI has still another 30 yrs before its everything and everywhere. You and everyone atm will be fine. Invest in people as well. The lost art of actually caring for your neighbor is almost dead.