Right now there is a two sides of AI "medal". Problem is both sides are weird.

All white collar job will be eliminated in 18 months

AI is bubble and it will burst very-very soon

Some numbers for first variant. White-collars are ~55% of all workforce in EU and USA. Current unemployment rate weighted average (EU+USA) ~5%. In case of 50% economical adoption of "all white collars jobs will be eliminated" we can expect 30% unemployment rate. As a reference Great Depression at it peak had a 25%. Multiply this to sheer amount of humans in 2026 vs 1930 to get a full picture.

Let dial down to more optimistic scenario. Jevons paradox adsorbed half of unemployments (not sure how for now, because transition from mid-developer to AI-orchestrator is not instant process). Plus after 10+ years of automatization, 50% is very optimistic adoption rate. So ~15% unemployment. It is still more than 2008 crisis. And if 2008 was a financial crisis (greed), possible crisis with 15% in 18-months will be structural.

Reverse side of medal "AI bubble burst"

Most of market grow in last two years was from Magnificent 7 and AI companies. Burst will eliminate inflated market costs, but infrastructure itself will remain. So AI remains, same as jobs. Collapse of startups GPT-wrappers, tech layoff (yes, in this scenario too), 100kk / year salary will vanish in AI field, bankruptcy of some hyperscalers but global job market will adsorb this shock more easily than structural collapse of workforce from first case.

Reallocation of capital from digital world to more classical and stable material things. From previous AI bubbles (it is not a first one) damage will be much bigger but will hit different areas. 1970 AI Winter was mostly academic and government spending with negligible macro-effect. Potential correction 2026-2027 will affect venture capital and public markets and we still have a byproducts of it - working LLMs, datacenters and infrastructure.

Happy(not) ending.

So is it worth to "burn Moscow to save the Empire"? I don't have an answer, I am just thinking out loud. There is a lot of variables, but almost none of them are positive. Middle ground between these two points also unpleasant. Surveillance state dystopia, increase in PII leaks, cyberpunk-style high-tech low-life, cognitive debt and possible cognitive decline.

We'll see.