We Think AI Is Useless. The Data Says Otherwise.

We Think AI Is Useless. The Data Says Otherwise. We're spending over $600 billion on AI this year. And what do we have to show for it? AI-generated slop. Deepfakes. Cat pictures in the style of Studio Ghibli. So yeah. I get why people think we're burning money on nothing. But here's the thing. That's not the full picture. Not even close. The AI You Don't See The AI that goes viral isn't the AI that's saving lives. It never is. Right now, Insilico Medicine is using generative AI to design cancer drugs from scratch. They went from zero to human clinical trials in 30 months. Traditional drug discovery? That takes 10 to 15 years and costs over $2 billion. Their AI-designed lung fibrosis drug, ISM001-055, is already showing positive results in Phase II trials. DeepMind's AlphaFold has mapped over 200 million protein structures. That's nearly every protein known to science. It won a Nobel Prize for it. Not a trending TikTok. A Nobel Prize. In 2025, MIT researchers used generative AI to invent new antibiotics against drug-resistant superbugs, including strains of gonorrhea and staph that we've been losing the fight against for decades. Google's FloodHub is giving 7-day flood warnings to communities in 80 countries. In Bihar, India, it cut medical costs by 30% because people could evacuate before the water came. In Kenya, an AI startup called Farmer Lifeline has detected over 1.4 million cases of crop pests and disease, reaching 42,000 smallholder farmers. Sixty-five percent of them are women. TurboTax's AI now auto-fills 90% of the most common tax forms. Last tax season alone, it saved Americans 6 million hours of manual data entry. Six million hours. That's time people got back to spend with their families, not squinting at 1099 forms. And yes, AI is even doing our laundry now. Figure AI's humanoid robot Figure 03 can fold clothes, wash dishes, and tidy your home. 1X's NEO robot ships to American homes in 2026 for $20,000, or $499 a month. A startup called Sunday Robotics trained their robot Memo using data from 500 real households, and it can now clear tables, load dishwashers, and pull espresso shots. None of this went viral. None of it. So Why Does It Feel Like AI Is Useless? Because gen AI's output is visual, shareable, and sits right in the middle of our daily communication. A Ghibli-style selfie gets 10 million views. An AI-designed antibiotic gets a journal citation. It isn't that we're only building useless AI. It's that useful AI doesn't make good content. Drug discovery doesn't have a share button. Flood prediction doesn't trend on X. A robot folding socks doesn't hit the same as a deepfake of a celebrity. The form factor of gen AI, text, image, video, is inherently viral. The form factor of useful AI, clinical trials, satellite data, protein folding, is inherently invisible. But Gen AI Isn't Entirely Useless Either Here's what the critics miss. A student in rural Sabah who can't afford tuition now has a personal tutor that speaks Malay, Mandarin, and English, available 24/7. A single mother in a kampung who's never met a lawyer can now ask ChatGPT how to file for child support. A first-generation college applicant who doesn't have parents who went to university can now learn how to write a scholarship essay, prep for interviews, and navigate the application system, all in their own language. Gen AI didn't just democratize content creation. It democratized intelligence. And for people who've never had access to privileged knowledge, like how to negotiate a salary, how to read a contract, how to get an internship at a Fortune 500, that's not useless. That's life-changing. Yes, there are misuses. Deepfakes. AI slop. Psychosis from people forming unhealthy attachments to chatbots. These are real problems. But the same argument applies to every powerful technology. Cars kill 1.3 million people a year. We didn't ban cars. We built seatbelts, speed limits, and traffic laws. The misuse of AI is a policy problem, not a technology problem. Why Big Tech and World Leaders Are Still Betting Billions Here's the part most people miss. The race isn't for today's gen AI. The Ghibli selfies. The AI slop. That's the byproduct. The race is for what comes next. The IEA estimates that widespread AI adoption could cut global carbon emissions by 1,400 million tonnes by 2035. That's three to four times more than all data centre emissions combined. DeepMind's wind energy optimisation has already boosted the economic value of renewable energy by 20%. AI is accelerating the discovery of new battery materials, new building materials, new carbon absorption methods. When Sam Altman says compute is the economy of the future, when countries like Saudi Arabia commit $100 billion to AI, when Canada pledges $2.4 billion and France commits €109 billion, they're not chasing chatbots. They're chasing the intermediary intelligence that could help us crack climate change, cure terminal illness, and end poverty at a scale that human brains alone simply cannot process. Is current gen AI the answer? No. But it's the path. The foundation models being built today are the infrastructure for tomorrow's breakthroughs. The same technology behind your Ghibli selfie is the same architecture being used to design cancer drugs and predict floods. The Real Question Are we ready for this? Probably not. Our regulations lag behind the technology. Our education systems haven't caught up. Our society is still processing the cultural shock of machines that can write, draw, and speak. But the answer isn't to stop building. It's to build better. Better policy. Better guardrails. Better distribution of AI's benefits so they don't just flow to Silicon Valley. So no, AI isn't useless. We're just looking at the wrong AI. The AI that matters doesn't trend. It treats. It predicts. It discovers. It teaches. And that AI? It's already here. We're just not paying attention.