Everyone is predicting AI now. Few people check the AI predictions from 30 years ago. I did. I went back 30 years and more. Books. Articles. Speeches. Ads. All published in 1996 or earlier. All public. I scored 48 AI predictions from 30 years ago.
The result is a clean split.
28 were accurate. 12 were partially accurate. 4 were wrong. 2 were too early. 2 are still pending.
But the split is not random. It follows a pattern. And the pattern matters for anyone building a brand today.
One group got the future right. They predicted partnership. Humans plus machines. Machines doing the routine work. Humans setting the goals. That is exactly how AI is used in 2026. The other group got the future wrong. They predicted timelines. Human level AI in a generation. Human level AI in three to eight years. They were out by 30 to 50 years.
Same era. Same brains. Opposite results. The lesson is simple. The people who predicted behaviour were right. The people who predicted dates were wrong.
Herbert Simon said it first. And he said it best. In 1971 he wrote that a wealth of information creates a poverty of attention.
Read that again. Slowly.
There was no feed in 1971. No social media. No programmatic ads. No creator economy. Simon still saw the whole game. Information would become abundant. Attention would become scarce. The scarce thing would become the business. That is the media business in 2026. In one sentence. Written 25 years before it existed.
Every marketer I know is living inside that sentence. Most have never read it.
Nicholas Negroponte wrote a book called Being Digital. It came out in January 1995. In it he described the Daily Me. A newspaper built for one person. You. Not a hundred pages for everyone. Ten pages for you.
The right ten pages. He went further. He imagined a control. Like a volume knob. Turn personalization up. Turn it down. Add some surprise back in. That knob still does not exist. The feed does.
Your LinkedIn feed is the Daily Me. Your TikTok feed is the Daily Me. Your news app is the Daily Me.
Negroponte also named the cost. Lose the shared edition and you lose shared reality. He flagged the filter bubble before the feed was built. He predicted the product. And he predicted the problem with the product. In the same chapter.
This one is my favourite. Because it came from advertising. In 1993 AT&T ran a campaign called You Will. Tom Selleck narrated. David Fincher directed.
The ads asked simple questions. Have you ever borrowed a book thousands of miles away? Have you ever taken a phone call on your wrist? Have you ever been rerouted around traffic by your car?
Then the answer. You will.
They showed e-books. GPS with live traffic. Video calls. Video on demand. Smartwatches. Tablet computing. Electronic tolls. Real time translation. Almost all of it arrived. Within 20 years. Now the hard part. AT&T built none of it. Other companies did. Mostly companies that did not exist when the ads ran. AT&T also missed mobile. The ads showed video calls from phone booths.
Hold both truths. The prediction was brilliant. The execution belonged to someone else. A prediction you do not build is a gift to your competitor.
In 1987 Apple made a concept video. It was called Knowledge Navigator. A professor picks up a tablet. A bow tied assistant appears on screen. The assistant understands plain language. It manages his schedule. It finds research. It joins a video call and shares documents. The video was set in September 2011.
Siri launched in October 2011. On the iPhone 4S.
Less than a month out. After 24 years. The idea was ready in 1987. The infrastructure was not. The chips were not. The network was not. The data was not. That gap is normal. Good ideas usually arrive early. Then they wait. The full version in that video is only arriving now. With the AI assistants of the last two years.
Now the misses. The misses are not stupid people. That is what makes them useful.
Herbert Simon and Allen Newell said in 1957 that a computer would be world chess champion within ten years. Deep Blue beat Garry Kasparov in 1997. Forty years later. Thirty years late. Simon also said machines would do any work a man can do within twenty years. That deadline was 1980. It has still not happened in 2026. Marvin Minsky went further. In 1967 he said artificial intelligence would be substantially solved within a generation.
In 1970 he said a machine with the general intelligence of an average human would arrive in three to eight years. We are still waiting. More than 50 years later. These were not outsiders. Simon won a Nobel Prize. Minsky helped found the field. Their overconfidence had a cost. It helped trigger the AI winters. Funding dried up. Credibility took years to rebuild.
Confidence is not evidence. A famous name is not a timeline.
Bill Gates published The Road Ahead in November 1995. He got a lot right. He described a wallet PC. Camera. Phone. GPS. Maps. Digital money. Biometrics. That is the smartphone. Twelve years before the iPhone. He described software agents that remember what you did before. That is the AI agent trend of today.
Then two big misses.
First, the internet. The 1995 hardback barely mentioned it. Microsoft was betting on its own network instead. A year later the paperback had to be rewritten. Whole chapters changed. About 20,000 words added.
Second, human like AI. Gates said it was not likely to happen in his lifetime. Large language models (LLM) arrived in the 2020s. Gates is alive. He has since changed his view. Publicly. The richest man in tech. With the best information in the world. Still missed the platform shift happening under his feet.
Proximity is not clarity. Sometimes it is the opposite.
David Ogilvy wrote 13 predictions for the future of advertising. They close his book Ogilvy on Advertising, published in 1983 and 1985. He got some things right. Global campaigns built at headquarters and adapted locally. Cheaper commercial production. Direct response moving to the center. He got print wrong. He expected a revival. He got billboards wrong. He expected them to disappear. They went digital and grew. He expected political campaigns to get more honest. No comment needed. But the real miss is an absence.
Zero digital predictions. No computers. No data. No targeting. No interactive advertising. Nothing.
The most famous adman of the century. Writing about the future of his own industry. And the biggest change in that industry is not in the list. Expertise has a blind spot. It is shaped like the present. Bill Bernbach saw the other risk. He warned the industry would worship technique instead of substance. Read that in 2026. In the age of dashboards and AI generated ads. He was describing us.
So ask better questions.
Second lesson. Own the future you predict.
AT&T predicted the future on national television. Then watched others build it. Apple drew its assistant in 1987. Then it shipped Siri, the iPad and the ecosystem around them. Slowly. But it shipped. A prediction is positioning. Building is proof. Most brands today are doing the AT&T version. Big AI statements. Keynotes. Vision films. Very little shipped.
Do the Apple version instead. Say less. Ship the small piece now. Then the next piece.
The future does not belong to whoever describes it best. It belongs to whoever is still building when it arrives.