Hope isn’t a strategy. Neither is denial

Abandon false hope, all ye who enter here

The Washington Post laid off one-third of its staff Wednesday, eliminating its sports section, several foreign bureaus and its books coverage in a widespread purge that represented a brutal blow to journalism and one of its most legendary brands… The Post’s executive editor, Matt Murray, called the move painful but necessary to put the outlet on stronger footing and to weather changes in technology and user habits (my emphasis)… The New York Times, which has been thriving in recent years, in large part due to investments in ancillary products such as games and its Wirecutter product recommendations. The Times has doubled its staff over the past decade.
AP News, 5/2/26

After spending all of 2025 warning AI was about to change everything, in December I begrudgingly conceded that, “Inconveniently for those of us who keep insisting AI will change everything, AI is yet to change very much”.

I noted that while pretty much everyone was talking about and experimenting with AI by late 2025, it was yet to have much of an observable real-world impact.

It was around this time several prominent AI boomers and doomers began to walk back some of their bolder predictions.

Daniel Kokotajlo felt the need to revise some of the timelines set out in his team’s attention-grabbing AI 2027 report. Having speculated that AI systems would be able to code autonomously by 2027, thereby fast-tracking Artificial Super Intelligence, Kokotajlo recently posted on X:

Things seem to be going somewhat slower than the AI 2027 scenario. Our timelines were longer than 2027 when we published and now they are a bit longer still; “around 2030, lots of uncertainty though” is what I say now.

Having once worked in the collapsing print media, I suspect many people who don’t want AI to turn their profession upside down are desperately grabbing at any ‘proof’ they can find that AI has been overhyped. Or at least that it will take several years before it has a profound impact on industries, economies and societies.

A few weeks is a long time in AI
But let’s consider some recent developments.

*Alphabet, Amazon, Meta and Microsoft spent US$360-$400 billion on AI in 2025 and are expected to spend US$600-$700 billion this year. That’s not too far off the US’s annual defence budget.

*After some new AI agent tools, notably Anthropic’s Claude Cowork, demonstrated they could handle complex workflows (i.e. research, CRM tasks, analytics) almost US$1 trillion was wiped from software stocks in recent weeks.*

*The Agentic AI Foundation, under the Linux Foundation, has now corralled 130 members, including heavy hitters such as OpenAI and Google, to create shared standards for AI agents, thus preventing closed ecosystems and keeping agent tools interoperable. (Like the W3C did for the web.)

*Speaking of Anthropic, strong investor demand recently saw its current capital raise go from $US10 billion to US$20 billion, based on a US$350 billion valuation. It already raised $13 billion last year.

After being launched by former OpenAI researcher Dario Amodei in 2021, Anthropic has become one of the world’s most highly valued private companies within half a decade.

So, while autonomous coding may still be some years off, other universe-denting developments are likely to arrive much sooner.

20/20 foresight beats 20/20 hindsight
At the risk of trying longer-term subscribers’ patience, let me refer once again to Dr Elisabeth Kübler-Ross’s five-stage model of how people almost invariably respond to the prospect of dramatic and unwelcome change – denial, anger, bargaining, depression and acceptance.

While transitioning through the first four stages seems to be unavoidable, those stuck in denial, anger, bargaining or depression often react sub-optimally, and sometimes self-destructively, to a looming tipping point.

Many journalists, along with their employers, failed to accept that the Internet was going to change everything. One of my favourite anecdotes about this is the response of a board member at a major Australian media organisation when informed that the rivers of gold advertising revenue that had long gushed into his company’s coffers would soon evaporate, with calamitous consequences.

Turning on the bearer of bad news, the corporate bigwig held up (some then hefty) dead-tree publications. In a classic anger-denial response, he then roared, “I don’t ever want anyone coming into this boardroom again and telling us that people will buy houses or cars, or look for a job, without [these papers].”

If you’re tempted to judge, consider how someone looking back from the perspective of, say, 2030 might view the attitudes to AI still common in early 2026.

Upsides and downsides
I once interviewed – aboard his impressive yacht, no less – someone who recognised technological change offered opportunities as well as posing risks.

He and his cofounders were prescient enough to realise people soon would be looking for jobs online rather than poring over table-cloth-sized pieces of paper covered with smudgy black ink. Seek, the job site he co-founded, currently has a market capitalisation of around A$6.5 (US$4.6) billion.

Much the same dynamic has played out with The New York Times and The Washington Post. One newspaper accepted the world was changing and changed along with it. The other, not so much.

The pertinent point is no individual or business can adapt to and, who knows, perhaps even profit from changed circumstances unless they concede circumstances have changed.

That sounds straightforward on paper, but is somewhat more challenging in practice. People fear change and will perform elaborate mental contortions to convince themselves it’s not really happening.

Or that if it is happening, that they will somehow remain unimpacted by it.

Gradually then suddenly
I don’t know how the AI transformation is going to play out. Anybody who tells you they do is either lying to themselves or you.

But as Hemingway observed, events tend to unfold gradually then suddenly. The Internet achieved mass penetration in the mid-to-late 1990s. It was years and arguably well over a decade before it started collapsing the legacy media’s business model.

But when the reckoning arrived, it hit hard and fast.

The world’s brightest minds are unable to predict when the ‘suddenly’ phase of the AI transformation will kick in. Maybe it won’t be until 2028 or 2030 or 2032.

But it just might be 2026. There’s little downside to contemplating that eventuality, but plenty to convincing yourself it’s entirely unthinkable.

* The new AI agents can autonomously perform many office workflows that SaaS businesses monetise on a per‑seat basis. That means they threaten to decouple software spending from human headcount, undermining SaaS’s “more employees = more licences = more revenue” golden goose. I suspect this won’t be the last we hear of human-headcount decoupling.