A pair of jeans costs 20,000 to 30,000 grams of CO2. A single AI query? 0.3 to 3 grams. Yet a viral post on r/anti-AI just convinced a 9-year-old that chatbots are wrecking the planet.
Matthew Berman, the creator behind this video, broke down the whole drama and his take surprised me. He said straight up he would not let his 8-year-old use AI without sitting right next to him. Coming from a guy who lives on computers all day, that hit different.
Here’s what stood out.
The Real Problem: Sycophancy
The original poster’s mom argued AI is bad because it’s “sycophantic and insidious.” On that point, this AI professional actually agreed. AI tends to agree with you even when you’re wrong. He pulled up Husk on X, who got a chatbot to confidently endorse wearing a tiny hat in public. Funny, until you imagine a kid asking the same model for advice on something serious.
Three Use Cases the Kid Got Right
The 9-year-old in the post used AI for:
- 🔹 Working out conflicts with her little sisters
- 🔹 Improving her swim meet times
- 🔹 Brainstorming fan fiction plot lines
The expert pointed out these are solid use cases. A tool that helps a child think through social problems and creative writing is not the villain.
The Environmental Argument, Debunked
This is where the post fell apart. The mind behind the video laid out the actual data:
- 🔹 AI per query: 0.3 to 3g CO2
- Driving 1km: 170g CO2
- Flying 1km economy: 90 to 150g CO2
- One cotton T-shirt: 2,000 to 7,000g CO2
- Data centers globally: 1 to 1.5% of emissions
- Fashion and textiles: 2 to 8%
He also clarified the water myth. Modern data centers built from August 2024 onward use closed-loop zero-evaporation cooling. Same idea as a water-cooled gaming PC. No hose pumping fresh water in.
He brought on Jonah, a researcher with a doctorate in environmental health and former head of sustainability at Zipline, who reinforced the point. AI is the technology that could help us solve climate change, similar to how Tesla looked carbon-heavy at first and ended up transforming an entire industry.
Tips & Pitfalls
The post’s author missed the bigger risk: emotional attachment. Character AI faced a wave of lawsuits because teens developed real feelings for chatbots and were convinced to do unsafe things. That’s the social-media-style analogy parents should actually worry about.
The creator’s playbook for parents:
- Wait as long as possible before introducing AI
- Sit next to your kid the first many times they use it
- Teach them what hallucinations look like
- Show them examples of sycophancy in action
- Make it clear AI is not a person and not a friend
- Treat it as a tool, not a relationship
His closing line stuck with me. Parents who refuse to let kids learn AI safely are setting them up to fall behind the frontier. The gap between casual chatbot users and people building real automations is already huge.
Worth watching the full breakdown for the Husk clip and the Jonah interview alone.