There is a meme that repeatedly pops up in my feeds that says something like “Adulting is just saying “it’ll calm down next week” repeatedly till you die”.
On this day in 2016, Davie Bowie died, which seemed to trigger a floodgate of celebrity deaths, but the type of old-school celebrities that got their status by some form of merit: Prince, Alan Rickman, Muhammad Ali, Gene Wilder, George Michael and Carrie Fisher all shuffled off this mortal coil a little sooner than people were ready to deal with in 2016. In parallel to the Grim Reaper seemingly mistaking the Hollywood Walk of Fame as his hit list for the year, the UK kicked itself in the proverbial knackers and voted to eject itself from the EU.
2016 was a shitty year, but it was incomparable to the years that followed, every year the world seemed to get worse and worse, global pandemics, horrific natural disasters, war, and a global political lurch towards fascism.
Global enshitification.
The tech industry seemed to follow the same pattern, culminating last year in what felt like the entire of Silicon Valley discarding the liberal costume it appears to have been hiding behind for the last few decades and revealing a desire to support whatever flavour of fascism they think will further their mission to destroy the world with technology.
I don’t think 2025 is going to be any better so I decided to jot down some of my predictions for 2025 (and/or soon after).
Disclaimer: I’m not a futurologist, and any valid predictions I have made about the global tech industry happen very slowly or with a much lower level of impact than I had previously predicted, so you can take this with a pinch of Private Frazer’s “We’re Doomed!” salt1.
Agent to Agent Decision Making.
I’m not sure why no one seems to be talking more about this, but we seem to be close to a much wider adoption of using AI Agent-to-Agent communication for decision-making. We’re already seeing the adoption of the practice of sending an AI agent to meetings people don’t feel are important enough for them to spend their time on to take notes and summarise the meeting.
It seems inevitable that very soon – if it hasn’t already happened – only AI agents will show up to a meeting. As these agents currently only work in a listening function, there will simply be silence – and in a lot of organisations – the recurring meeting will likely continue with only the presence of AI agents every week until the original organiser leaves the company and their calendar is whipped.
But what if you can send a more active agent to the meeting? The main purpose of meetings is to align a topic against a group of people’s opinions (fact-based, or otherwise). If you could train an AI agent with your personal opinions on professional topics, you could send it to answer any questions and agree with others on the correct compromises.
If everyone does the same, just connecting our Agents to make a collective decision seems a possible reality, at least in theory. It’s only a matter of time before someone tries it.
This will significantly shift the conversation around which jobs are likely to be impacted by the AI revolution and potentially create a marketplace for AI agent personas trained on successful knowledge experts.
The Great AI Price Hike.
Just like every SaaS vertical over the past few years, companies offering AI services are all running at a VC-funded Loss Leader pricing strategy to help them get traction in the market.
Machine Learning and other AI technologies are incredibly processor-intensive, which comes with high costs. There are two outcomes that happen here: either the computational costs drop quickly enough to enable the costs of AI services to remain at their current levels before the companies run out of money, or they will hike their prices once individuals and companies have become dependent on their services.
If history (cloud computing and software observability services) has taught us anything, it will be somewhere along this scale, but more than likely, a significant amount of costs will be transferred to the end user.
”Does Globalisation Work?” or “Local Products For Local People”
Historically, communities have developed in fairly isolated environments. Developing their own belief systems and moral codes separately from other communities. Cheap air travel and the internet threw all of these communities into the same space in a large multicultural melting pot of chaos, causing conflicts and cultural challenges. Whether you agree or not, the prevalent argument on this topic seems to be that it isn’t working, and we need to reverse this trend.
This has implications for tech, which has been using easy access to global markets as a big factor in financial projections. We are already seeing the topic arise with the USA’s potential impending ban of TikTok and the reluctance of non-EU companies to launch AI products within the EU (Apple Intelligence being a great example).
I think this will continue to have an impact through 2025, pushing companies to think much more about targeting their products to a market they understand and also narrowing their hiring practices to focus on more political concepts of “cultural fit”.
The Rise of the Neo Luddites and Artisan Tech
Generative AI is churning out a LOT of crap. There are already studies popping up that suggest generative AI’s ability to speed up the productivity of engineers and other knowledge workers are causing a slowdown in the overall company’s productivity. In Engineering specifically, the model that “more code = more problems” seems to be surfacing again and again.
I think people will start to think about simplicity in the solutions they provide, avoiding complicated technical solutions in favour of more proven solutions that can be quickly created without high levels of complexity.
I predict that the quality of engineering solutions vs quantity will be a big topic in 2025 and beyond.
It will become evident that the answer to the question of how we “do more with less” isn’t generative AI but instead just by doing less better.
The (Very) Slow Tech Industry Re-org and a Further Lack of “Pandering”
This should have happened already, but I think the VC funding crash slowed it down; but I think 2025 will finally be the year we stop talking about “Office vs Hyrid vs Remote.” We will start to have an industry-wide re-org where companies that are trying to force remote people into the office (and visa versa) will start to fail, and companies that have the right types of people for their workplace philosophy will succeed.
The failing companies that have the wrong people will slowly start to cycle their workforce (forcefully or organically), and we’ll eventually end up with more of the people working in an environment that fits them.
However, this won’t be a soft handholding process where every engineer’s wishes and desires are pandered for to make sure they’re happy. People will be fired, departments will be made redundant and engineers will have to dig deep and work hard to get the jobs that they really want.
Maybe the spirit of Private Fraser is in the driving seat today, but my recommendation for 2025 is to buckle up, it’s not going to be an enjoyable ride.
The jokes always used to rotate around “Who’s still alive from the cast of Dad’s Army?”.. they will soon be “Who’s still alive that will get a Dad’s Army reference”.
Leave a Reply