You know what they say: predictions are like resurgent nationalist movements. Everyone’s got one.

I missed the deadline for Nieman Lab’s always-excellent Predictions for Journalism this year, so I thought I’d share a few more bite-sized predictions about various topics I’ve written over the last year. Every prediction says more about the person making it than about the actual future; please take these in that light. I am not a soothsayer, but boy, do I have opinions.

Here are some of them:

The AI industry will continue to orient itself around its definition of AGI, regardless of its harms.

OpenAI and Microsoft’s definition of artificial general intelligence is not what you might suspect: they define it as the point where AI systems can generate at least $100 billion in profits. Given that the industry is losing billions of dollars hand over fist today, there’s a long way to go.

Closing that gap means selling in lots of different places, but the most lucrative are going to be deeper partnerships with mass-market systems, government, and military applications. For all of OpenAI’s talk about not creating AI that will make us extinct through its intelligence, I predict it and companies like it will take firmer steps towards assisting companies who might kill us through more prosaic means.

AI vendors may also look at ways to reduce the cost of sanitizing and tagging its input data — currently often outsourced overseas. They may, for example, consider using prison labor, taking cues from Finland, which has engaged in the practice for years.

Publishers will pivot to AI, with predictable results.

Lured by up-front payouts and a carefully-cultivated (and heavily paid-for) sense that they’re missing out if they’re not participating, many news publishers will be all-in on AI. It will be to their detriment.

Publishers with low-volume qualitative output will mistakenly think that their high-quality stories are more valuable to AI vendors, fundamentally misunderstanding how training data is acquired and used. They will not see the ongoing licensing premiums for their content that they might hope for.

Publishers with high-volume output will allow their stories to be used as training data. They will find that ongoing revenue suffers as a result and that those payments only temporarily addressed a downward funding trend that will continue apace.

Only the publishers who treat AI as a side issue and continue to address their fundamental value to their readers and communities will succeed.

The United States will not create a Bitcoin reserve.

Despite calls and even a pledge to the contrary, President Trump will not follow through with creating any kind of crypto reserve or an intentional stockpile of Bitcoin. It’s simply not in his interests: the US dollar is not just a currency but a global network of power and influence that he can leverage to his advantage.

But don’t rejoice quite yet, crypto-skeptics. Instead of stockpiling existing, independent cryptocurrencies, he might plausibly create a new coin with US interests in mind and with the official seal of a government endorsement, with partners drawn from his existing network. (USDC, the prevailing dollar-backed stablecoin, is issued by Circle, a private company. This would be a replacement.) The result would almost certainly be more profit for his own private interests and that of his friends, particularly as he could incentivize traditional American banks to support it as a transfer mechanism.

Threads will implement full ActivityPub integration but continue to struggle to release it in the EU.

Confounding its skeptics, Threads will release full end-to-end support for the ActivityPub specification that allows it to act as one cohesive social network with Mastodon, among other platforms. The immediate effect will be a change of the center of gravity in the Fediverse: rather than Threads being seen to integrate with Mastodon, Mastodon and every Fediverse platform will be seen as Threads-compatible. (Mastodon et al will continue to support smaller communities with specific needs; Threads will be the mass market platform on the network.)

Because of the way data is federated between systems in ActivityPub, and because of Meta’s data commitments as a large platform owner, this compatibility will not launch in the EU without major changes to the experience. Meta will endeavor to work with the authors of ActivityPub to make it easier to comply with EU data restrictions, but may be seen as trying to exert undue influence over the protocol by some in the community.

Some social media platforms will relocate from the US.

In an effort to maintain independence and avoid complying with restrictions to Section 230 and an uptick in government subpoenas under the Trump administration, some social media platforms will move their headquarters to countries that allow them to maintain more independence.

Neutral Switzerland will be a favorite. Because of a local requirement to have some Swiss ownership of countries located there, some founders will seek to go through its notoriously difficult naturalization process; there will also be an influx of repatriated Swiss tech entrepreneurs who see an opportunity in helping out.

TikTok will continue to operate, but will need to take it to the Supreme Court.

The law banning TikTok goes into effect on January 19, one day before the inauguration of the new President. It cannot comply. It’s likely, therefore, that it will take up the case and bring it to the Supreme Court. The Court may then decide that the law was written with punishing a single target in mind (TikTok alone), without a preceding trial for the claimed crimes, and could repeal it on that basis.

Bird flu will be a thing.

California has already declared a state of emergency because of its spread in cattle, and the virus has already mutated in human hosts to become more infectious. 66 people have died from it at the time of writing. On the prediction markets, the probability of a million cases by the end of the year is soaring.

Whether this becomes a global pandemic like COVID-19 will be up to governments to respond. Given the US government that will be in power when this does, inevitably, become a thing, I’ll leave it up to the reader to decide whether the response will be science-based and adequately up to the challenge.

Long-form fiction will (continue to) rise.

A lot of ink has been spilled about the death of books. Elle Griffin’s piece No one buys books has been particularly influential. It’s also not a complete picture.

It’s absolutely true that the big publishing houses are consolidating and that there are fewer opportunities to be published by them if you don’t have an existing community. But there’s a long tail of smaller publishing houses, and self-publishing has become more than a cottage industry. The latter isn’t just hacks banging out AI-written non-fiction self-help books; there are many, many authors building genuinely great careers on their own terms. They’re not Stephen King millionaires, but they’re making a great living — particularly in genres like dark romance that big publishing houses might be less excited to touch.

In a world that is going to feel a bit more adverse (see my other predictions above), independent, interesting fiction that speaks to the needs of its audience will both find that audience and do well with it. In turn, the continued rise of ereaders will make the relative lack of placement in bookstores for those titles almost irrelevant. Fiction is undergoing the classic disruption story; it’s not dying at all.

This disruption will accelerate in 2025. There’s even an opportunity to do for long-form fiction what Substack did for newsletters, and I’d bet that someone will take it. Even without such a platform, the Kindle Direct Publishing program and services like IngramSpark (together with sales support from the likes of BookBub etc) will allow the market to continue to grow.

Unions movements will continue to grow, particularly for knowledge workers. Whether they’ll win is up in the air.

The labor movement continues to gain strength, and unions have historically high support, although actual union membership remains incredibly low. The first trend is likely to continue, particularly as AI continues to threaten the livelihoods of knowledge workers, and as the Trump administration emboldens employers to roll back benefits and DEI initiatives: they will attempt to unionize in greater numbers, with more ferocity, and more interruptions to work while they negotiate for stronger protections.

Will they win? I don’t know. Union contract negotiations can take years, so it’s hard to say what the outcome will be. If they do win, the outcome will be higher wages, stronger benefits, and better working conditions for employees. (That’s what unions do.) But historically, knowledge worker unions have had a hard time convincing colleagues to sign up; see the Alphabet Workers Union, whose membership is a tiny fraction of Alphabet’s total employment base.

What did I miss? What did I get wrong?

Those are some of my predictions for 2025. What are yours? Where do you disagree? I’d love to hear from you.