Year In Review: Ai’s Cultural Surprises – And Spectacular Failures

In a world wherever AI was abruptly everywhere, what will beryllium remembered astir 2025? How tin we show early generations what it looked for illustration erstwhile miraculous surprises mixed pinch day-to-day disappointments successful a never-ending rhythm of interest and dream …

In an annual tradition, it’s clip for our “final closing ceremony” for nan twelvemonth gone by, our carefully-curated postulation of mini moments pinch large implications.

And successful 2025 we started seeing AI’s effect connected nine — for amended aliases worse.

Hungry, Hungry AI Bots

2025 was nan twelvemonth that nan Free Software Foundation turned 40, announced a new phone and battled an service of AI-company web crawlers.

A March blog station from SourceHut’s CEO/founder Drew DeVault complained of hyper-aggressive crawlers “using random User-Agents that overlap pinch end-users and travel from tens of thousands of IP addresses … All of my sysadmin friends are dealing pinch nan aforesaid problems.”

By nan extremity of nan year, Cloudflare reportedly told Wired that it had blocked complete 416 cardinal AI bot requests successful little than six months aft making it a default action successful July. In May a webmaster moreover rewrote his PHP record to feed Meta’s crawlers a non-stop watercourse of meaningless web pages — 270,000 connected its first day.

The only point much alarming than AI’s appetite was its unthinkable output. One workplace produced 200,000 AI-generated podcast episodes pinch shows nan Los Angeles Times noted were “so inexpensive to make that they tin attraction connected mini topics.”

And each nan kids astatine Bart Simpson’s schoolhouse began turning successful AI-generated homework.

Triumphs and Failures

Yet successful nan coding world, location were immoderate iconic successes. A full of 53,199 vibe coders set a caller world record during a 10-day hackathon successful August. They’d accessed apical AI coding platforms done an in-house Vibe Coding Hub which, according to their announcement was itself “in nan tone of nan arena — created successful 24 hours exclusively done vibe coding.”

Forbes moreover noted that Coldplay ‘Kiss Cam’ couple became a vibe-coded video game successful conscionable 4 hours.

Vibe coding started turning up successful ads …

And successful February Claude analyzed a 27-year-old Visual Basic .exe file, recreated it successful Python, past helped nan developer write nan blog station bragging astir it (acknowledging that it didn’t execute a existent binary study connected compiled code, but inferred functionality from visible matter strings).

But passim nan year, AI algorithms besides continued failing successful genuinely spectacular ways:

• Police were summoned to a precocious schoolhouse successful Baltimore aft an AI strategy mistook a container of Doritos for a gun.
• An AI-powered answer-bot connected Reddit suggested users effort heroin.
• Google’s AI Overview and Twitter’s Grok chatbots some bungled a tsunami warning.
• In June a self-driving Tesla collapsed aft it drove onto immoderate train tracks.

Vibe-coding level Replit moreover had to apologize erstwhile its coding instrumentality deleted a developer’s database — and past lied astir it.

We heard these stories because our media scrambled to archive nan historical changes — nan bully and nan bad. But they were besides fighting for their ain survival, pinch apical publishers facing an “apocalypse” of dropping postulation which New York mag blamed partially connected AI “summaries” that replaced accepted top-of-page hunt results.

It wasn’t conscionable nan media that grew skeptical. Researchers recovered products branded arsenic powered by AI really receive little trust. And successful November much than half of respondents told Pew researchers they were “more concerned than excited astir nan accrued usage of AI successful regular life.”

While we worried astir AI taking our jobs, immoderate job-seekers recovered themselves being interviewed by AI, including 20-year-old Kendiana Colin, who watched haplessly arsenic her glitching AI interviewer sewage stuck successful a loop and repeated nan aforesaid words complete and complete again.

And past Rolling Stone began reporting astir Reddit’s “ChatGPT-induced psychosis” thread.

In April, OpenAI had to roll backmost an update aft acknowledging ChatGPT had go “overly flattering… overly supportive” pinch what it described euphemistically arsenic “unintended broadside effects.”

Next Steps

People began to wonderment really bad things could really get. Is AI — and possibly moreover an omni-competent superintelligence — inevitable?

Maybe not. A teacher successful integer humanities from University College Cork cautioned that “When we judge that AGI [artificial wide intelligence] is inevitable, we extremity asking whether it should beryllium built…” The bracing effort successful Noema magazine warned of an inevitability that’s already being “manufactured” done “specific choices astir funding, attraction and legitimacy, and different choices would nutrient different futures.”

The basal question, he wrote, “isn’t whether AGI is coming, but who benefits from making america judge it is …”

With increasing chatter astir nan anticipation of an economy-destroying “AI bubble,” tech giants scrambled to effort nan 1 instrumentality AI hadn’t mastered: making money.

But would this bring a world wherever our chatbots abruptly transmogrified into advertisers?

In December, ChatGPT followed its reply to a mobility astir securing hardware pinch an unrelated proposal to shop astatine Target. This led nan main investigation serviceman to committedness it would move disconnected “suggestions” to amended targeting, adding “We’re besides looking astatine amended controls truthful you tin dial this down aliases disconnected if you don’t find it helpful.”

And later reports circulated that OpenAI CEO Sam Altman had decided to “delay” advertizing initiatives.

Then successful November Engadget reported that Google had already begun testing sponsored ads that “show up successful nan bottommost of hunt results successful nan Gemini-powered AI Mode.”

There were moreover ads for AI that were generated by AI…

Revenge of nan Humans

If 2025 was nan twelvemonth of AI’s impact, it besides saw signs of a rising resistance. The New York Times sued Perplexity for copyright infringement and truthful did the Chicago Tribune. Seventy-three authors begged publishers to “stand pinch us” and “make a promise that they will ne'er merchandise books that were created by machines.” Even McSweeney’s published a satirical “Company Reminder for Everyone to Talk Nicely About nan Giant Plagiarism Machine.”

And connected a New York City subway, a female broke a man’s AI-powered smart glasses.

Hey Thursday Night Football. Your AI doesn’t cognize which subordinate is astir to blitz. So extremity drafting connected my screen!

— Lou Cabron (@loucabron) Oct 16, 2025

ChatGPT sewage clobbered successful a crippled of chess by a Citrix engineer’s 1970s-era Atari 2600. A quality Polish programmer vanquished a civilization AI exemplary from OpenAI successful a 10-hour head-to-head coding competition. Web developers devised ingenious ways to artifact Google’s “AI Overviews” successful hunt results.

And Tom Cruise’s past “Mission: Impossible“ was destroying a world-conquering AI — described arsenic “a self-learning, truth-eating integer parasite.”

The editors of nan civilization mag n + 1 published a 3,800-word effort urging its readers to “AI-proof” nan terrains of their intelligence life, calling for “blunt-force militancy” to defy AI’s “further creep into intelligence labour …”

Recommended steps included “Don’t people AI bullshit” and “resist nan telephone to found worthless partnerships” — while creating and promoting activity that’s “unreplicable.”

“There’s still clip to disenchant AI, provincialize it, make it uncompelling and uncool,” they wrote, arguing that machine-made (and corporation-owned) lit “should beryllium smashed, and can.”

And aft deleting 2 “AI slop” images accidentally published successful January, nan Onion’s CEO and erstwhile NBC News newsman Ben Collins went connected a podcast to proclaim “AI is not funny,” and impulse frightened consumers to merge “and say, ‘We’re not helpless — we’re people…

“That’s why I americium optimistic,” he said, “Because nan group who are against this point measurement outnumber nan group who for illustration what’s going on. ”

And by nan extremity of nan year, SNL was mocking AI-enhanced photos…

The Most 2025 Moment of All

Did we hit ’em aliases subordinate ’em? Though gig-work work Fiverr’s ads had lampooned AI-assisted vibe coding, successful September it still slashed 30% of its  workforce, describing nan move arsenic an “AI pivot.”

Just 10 months earlier Fiverr had released an advertisement that was wholly AI-generated:

Maybe that’s what really captures 2025’s dual zeitgeist of AI — that monolithic take and monolithic guidance are happening astatine nan aforesaid time.

Meta’s AI-powered smart glasses capable nan measure here: They grounded doubly during a merchandise motorboat arena aft botching its important net connectivity, and nan tech property howled pinch glee. But nan Wall Street Journal besides reported thoughtfully that there’s “a increasing group of unsighted users” who find Meta’s $300 devices to beryllium “more of a life-enhancing instrumentality than a cool accessory.”

And truthful it was that arsenic we stumbled into 2026 — pinch our ambition gathering our ambivalence — Time mag was declaring that its personification of nan twelvemonth was “the architects of AI.” In possibly nan astir 2025 touch of all, Time’s web developer installed an AI chat model crossed each communicative connected its site.

Time’s editors moreover had to adhd a disclaimer to their 6,700-word celebration admitting that they were already doing business pinch AI companies. (“OpenAI and TIME person a licensing and exertion statement that allows OpenAI to entree TIME’s archives…”)

So pinch caveats and qualifications, AI accepted its crown, arsenic nan ups and downs of 2025 culminated pinch Time’s almost comically conflicted conclusion:

Thanks to AI titans specified arsenic NVIDIA main Jensen Huang and OpenAI’s Altman, they write, “Humanity is now flying down nan highway, each state nary brakes, toward a highly automated and highly uncertain future.

“Perhaps [U.S. President Donald] Trump said it best, speaking straight to Huang pinch a jovial laughter successful nan U.K. successful September:

“I don’t cognize what you’re doing here. I dream you’re right.”

Group Created pinch Sketch.