Author: Eric Worrall

What do you tell demoralised kids who think AI has eaten their future?

AI fears grip graduates worldwide as future of work feels uncertain

TUESDAY, JUNE 16, 2026
|AKARIN VIBULTANGMAN

• Recent university graduates at several US commencements have booed speakers for promoting AI, demonstrating a visible and growing anxiety about the technology’s impact on their future careers.
• This fear is intensified by a difficult entry-level job market, with graduates worried that AI will automate or eliminate the very junior roles like research, data support, and content production that serve as the first step on a career ladder.
• The graduates’ concern stems from their familiarity with AI; having used tools like ChatGPT in their studies, they understand its capabilities and fear employers will use the same technology to reduce hiring.
• Recent surveys confirm this sentiment, with a Harvard Youth Poll finding that 59% of young Americans view AI as a threat to their job prospects, and a Gallup poll showing increased anger and anxiety about AI among Gen Z.

…

Read more: https://www.nationthailand.com/news/world/40067476

Junior jobs in a range of industries are under threat.

What the Data Actually Shows About AI and Junior Developer Employment Decline

AUTHOR James A. Wondrasek

The debate about whether AI is displacing junior developers has moved past the speculation stage. We have data now. But the data needs careful reading, because it does not all point in the same direction.

The strongest evidence comes from the Stanford Digital Economy Lab. …

…

What does the data actually say about junior developer employment?

Workers aged 22-25 in the most AI-exposed occupations experienced a 16% relative employment decline compared to the least AI-exposed occupations. That comes from the Stanford/ADP study. In absolute terms, employment for this age group has fallen roughly 20% from its peak.

Here is where it gets interesting. Workers aged 35-49 in the same AI-exposed occupations saw 6-9% employment growth over the same period. The decline is not hitting everyone. It is hitting the youngest workers in roles where AI does the most work.

…

Why are junior developers affected while experienced engineers are not?

It comes down to what kinds of knowledge AI can replace. This is also why AI is compressing teams rather than replacing programmers — the compression is selective.

…

Read more: https://www.softwareseni.com/what-the-data-actually-shows-about-ai-and-junior-developer-employment-decline/

Clearly AI can replace junior software developers, at least to an extent. But senior jobs don’t seem as heavily affected, though this is cold comfort to graduates who expect to enter the industry as a junior.

What about other professions?

Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence

Revised August 12 2026

Using a sample of high-frequency administrative payroll data from ADP covering millions of U.S. workers through June 2026, we document six facts about the labor market following the widespread adoption of generative AI.

1. We find no evidence of widespread, economy-wide job displacement.
2. However, employment of young workers (ages 22–25) in AI-exposed occupations now stands 19% below where it would be had it kept pace with that of their less-exposed peers; experienced workers show no comparable gap.
3. This divergence has widened steadily since we first documented it in August 2025.
4. It operates primarily through reduced hiring of young workers rather than increased separations.
5. Declines are concentrated in occupations where AI usage primarily substitutes for human tasks; where usage primarily complements workers, employment is flat or rising, especially for experienced workers
6. Adjustment is occurring through employment rather than base compensation.

…

Read more: https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/

There’s a hint – the impact of AI on the economy is not evenly distributed. Some jobs are more resistant than others.

What Young Workers Are Doing to AI-Proof Themselves

They’ve got their whole careers ahead of them, and they’re navigating a technology with a still-uncertain impact

By Rachel Wolfe and Te-Ping Chen
March 22, 2026 at 5:30 am ET

Jackson Curtis planned to spend his career in insurance, where he’s worked the last 3½ years. Instead, the 28-year-old is now pursuing an abrupt shift: becoming a full-time firefighter.

…

“Even if they can come up with a way to utilize AI to fight fires, people are always going to want that empathy from an actual human who actually cares during a moment of crisis,” Curtis said. 

…

When Ryder Paredes, 22, began studying computer science as an undergraduate three years ago, “AI was in its infancy, it wasn’t very intelligent,” the Montclair, N.J., native said. But by last year, AI’s abilities had improved so much Paredes feared he wouldn’t be employable. 

He dropped out of college last year and is now in trade school studying to be an electrician. He isn’t alone: Enrollment at vocational-focused community colleges has boomed in recent years, growing by nearly 20% since 2020, according to National Student Clearinghouse data.

…

Read more: https://www.wsj.com/economy/jobs/ai-jobs-young-people-careers-14282284

The people quoted in the WSJ article, Jackson and Ryder, are absolutely right. Jobs which involve interaction with unpredictable real world situations are highly resistant to AI.

The reason is simple – it is far easier to train an AI to deal intelligently with the electronic world than with the real world.

Training an AI to write code – there is a huge amount of information about how to write code online, including large online howto repositories which can be fed straight into an AI as training data. Code writing skill is also easy to evaluate – ask your trainee AI to create some code which does something complicated, over and over, then run it for a few months if necessary until it crashes. When it crashes, have the AI work out why and your AI also develops fault diagnosis skills.

All this AI training for software development happens at electronic speeds. AI competing with junior software developers makes software development an AI replaces the human profession. Thanks to AI, it is getting tougher to get your foot in the door as a junior software developer, into an industry which was already tough to get into.

Training an AI to complete a plumbing job – good luck with that. Your training ground isn’t a big computer server quietly tucked away in some data center, your training ground is the real world. Messy, non standard, amateur, pissing down with rain or snow, angry householders demanding you fix it yesterday. Putting an untrained robot, or one trained only on plumbing simulations, into that messy real-world environment, where it’s 99.9% likely to make things worse until it has learned from a million real jobs what not to do, is a far harder proposition than replacing a software developer. And don’t even mention gas fitting, where the consequences and legal repercussions of doing a bad job can involve lots of dead people and big lawsuits.

Firefighting is an interesting intermediate case. There are regular “firefighting robot” competitions, in which a little mouse size robot runs around a maze to put out the fire, with prizes for performance. There are robots which inspect dangerous environments in firefighting situations to reduce the risk to human life. But how quickly this advances to actually replacing humans, I suspect in firefighting humans will have the upper hand for quite some time. For now at least firefighting is a robot helps the human profession rather than a robot replaces the human profession.

Not all blue collar workers are so lucky – a lot of skilled metal machinist jobs were erased by CNC controlled metal working machines in the 1990s. The machinists who survived best worked at the unpredictable end of the trade: small one-off jobs and bespoke work. Unpredictable metalwork-adjacent trades, such as automobile mechanic, continue to thrive. Unpredictable = difficult to automate.

So what if you don’t like getting your hands dirty like a plumber, or fixing automobiles, or hard charging into high risk situations like a firefighter?

The good news is there are plenty of other options to choose from. Just as metal work adjacent professions such as automobile mechanic survived the skilled metal machinist bloodbath, so there are some professional occupations which are adjacent to vulnerable professions like software development, but which are far more resistant to AI incursion.

One of my favourite software development adjacent professions which is resistant to an AI takeover is data science.

Some data science jobs are predictable and therefore easy to automate. But people who work at the messy end of data science, teasing meaning out of poor quality and corrupted real world data, helping AI overcome deficits in data which it cannot overcome unassisted, they’re likely to do just fine.

So if a young person is interested in a white collar technical career, and they are good at mathematics and really want to beat the AI, give them a nudge towards data science, or tell them to search for careers which are messy, which involve unpredictable elements which are difficult to automate.

My point is, for all of AI’s transformative impact on the global economy, there are still plenty of unpredictable niche professions where human beings will continue to dominate for the foreseeable future. It is just that as with all great societal upheavals, you can’t just mindlessly apply old world solutions and expect everything to turn out fine. You need to use your brain, and find new paths through a future career landscape which is very different to what your parents faced.

I love WordPress but there is a problem – WordPress search sucks.

What if you could replace WordPress search with something better?

Well you don’t have to wonder anymore. As an experiment I’ve done just that. I created a custom plugin for searching this site which sends the search request to an AI, to produce a list of pages which match the request.

The AI I’m using is Pinecone – a vector database engine.

Vector database engines try to convert text into meaning. If you type a sentence like “I am a cat”, the Vector AI converts that sentence into a series of numbers, numbers which help the AI measure the distance between the meaning of what you just typed, and a set of searchable text documents.

How it does that conversion – that’s something I’m still researching.

But the point is, by using a vector engine like Pinecone, you can perform rapid comparisons between diverse chunks of text, to produce a high quality list of what text corresponds to your search term.

Unfortunately Pinecone doesn’t support a PHP API. PHP is the language WordPress is written in. But that’s OK because I’m a software developer, and Pinecone published enough information about their APIs so I figured it out.

Give it a try – try entering “What is internet of things”, or some other search term into the search box on this website. As this is an experiment I’m kind of only using free tier, so you guys trying it out might exhaust the quota, but while it lasts it will be an interesting experience – far, far superior to normal wordpress search.

Does improved search on your website sound like something which would interest you? Should I package this up as a free or paid for plugin? Let me know your thoughts – and let me know if you are as impressed by the improvement to wordpress’ normal search as I am.

Have you ever wondered how vibe coding works? The following video demonstrates the process from start to finish of developing a simple game.

And here is the game:

I used Claude to develop this game, but any advanced AI would do.

The entire process took half an hour. If I had coded this game from scratch, it would have taken half a day.

Before you rush out thinking the world doesn’t need programmers anymore, there is a catch. I used my programming skill to structure the prompts. My first attempt to build this game didn’t work, the AI got stuck in a loop, churning up more and more credits while not doing anything – I needed my programming skill to understand why. Vibe coding works best for programmers, we know what prompts to write.

Eventually of course this will change – AI will become sufficiently advanced that anyone will be able to code as well as any programmer.

As the project progressed, the AI became more and more adept at figuring out what I wanted to do next, from memory I accepted two of the AI’s suggestions verbatim, without making any changes.

I also expect Vibe Production or Vibe product development to soon become a thing – creating physical objects rather than software. The ultimate drop shipping experience will soon be a reality – instead of scouring the internet for something to sell, you will be able to create an entirely new product out of your imagination, including promo material, and have the prototype in the mail in time for your morning coffee. This is already starting to happen – Meshy.ai allows 3D models to be created using vibe, models which can be fed straight into a desktop 3D printer. I expect to see this start to go mainstream from 2027, 2028 at the latest.

Welcome to the future.

The neural network below wants to learn how to drive. Press “Start AI Demo” to see a real neural network trying to solve a problem.

How does AI work?

Have you ever heard the saying, that a monkey hitting keys at random on a typewriter keyboard for an infinite amount of time will eventually type the complete works of William Shakespeare?

Of course, you would need to wait a very long time for your theatre script.

But what if you could take all the bits the monkey got right, and teach the monkey to focus and build on the good parts?

OK, this is probably stretching the metaphor to breaking point, but in a very real sense this is how AI works.

AI starts with a set of completely random attempts to solve a problem.

Just by chance some of those random solution attempts will be slightly better than the others in the set. AI builds on those slight improvements, and uses a process of iterative improvement to finally produce a solution.

This is why AI uses so much electricity and computing power – you have to sift a lot of poor quality attempted solutions, to find the one solution which answers your requirements. Sifting all those failures takes a lot of computing power.

Should I include AI in my next solution?

Maybe. AI is worth considering when conventional programming techniques are struggling to solve a problem. If you wanted to build an app which say recognises objects in photographs, using conventional programming to home in on the features of the object which help identify it would be extremely challenging. But an AI might be able to zero in on those identifying features in hours, a task which might take a human programmer months to solve, if they ever solve it.

What are the disadvantages of using AI?

AI is expensive to develop, and expensive to run. It is very difficult to estimate how long it will take to develop an AI, because AI isn’t amenable to usual quoting techniques – breaking down the problem into small steps, then analysing the cost of those steps.

If you would like to discuss whether AI can help your next project, Desirable Apps is happy to assist. But be aware, a serious AI project requires a serious budget, far more so than regular software development.