It did indeed. So I asked DeepSeek to compare AI training to 12 everyday products that consume more and are basically useless. Mind, that list can go into hundreds of thousands …
Here’s DeepSeek:
AI training does use resources. Training a large model can consume megawatt-hours of electricity and millions of liters of water for cooling. That deserves scrutiny and efficiency improvements.
But scale and purpose matter. That same AI model can then be used millions of times to:
· Accelerate medical research
· Reduce energy use in logistics
· Optimize building HVAC systems
· Help write code for renewable energy management
Meanwhile, the following products consume comparable or greater resources and produce zero lasting utility—often ending up in a drawer or landfill within months.
If your concern is resource waste, these should be your first target.
12 Far More Wasteful & Useless Things (vs. AI training)
Examples of Wasteful Items
Fidget Spinners (Peak Overproduction)
• Description: Billions of dollars of plastic, metal, and lithium batteries (for light-up versions) invested in a toy with no cognitive or fitness benefit.
• Usage Pattern: Most were used for days, then landfilled.
• Environmental Impact: The embodied energy exceeds training multiple small AI models.
Aerosol Hairspray
• Composition: Each can contains propellants (often greenhouse gases), alcohol, polymers, and a steel/aluminum container.
• Functionality: Provides temporary hold that gravity undoes in hours.
• Environmental Impact: Global production uses millions of tons of CO₂-equivalent emissions annually—far more than training all Large Language Models (LLMs) to date.
Seasonal Single-Use Decor
• Examples: Mass-produced plastic jack-o’-lanterns, Easter grass, tinsel.
• Materials: Virgin PET and PVC.
• Environmental Impact: Global oil use exceeds that of every large AI model ever trained.
Free Promotional USB Sticks (4 GB)
• Production Volume: Billions manufactured yearly.
• Usage Pattern: Most used once then lost.
• Components: Each contains a controller chip, NAND flash, rare-earth solder.
• Comparison to AI: AI at least retains its weights.
Keurig K-Cup “Innovation” Displays
• Description: Retail endcaps with 200+ pod varieties, refrigerated, lit 24/7.
• Environmental Impact: Energy to display one store’s pods for a month exceeds training BERT or GPT-2.
Cable Tie “Multi-Packs” from Dollar Stores
• Quantity: 500 nylon ties per bag.
• Usage Pattern: 90% never used.
• Production: Nylon production is energy-intensive (high heat, caprolactam synthesis).
• Comparison to AI: AI training doesn’t permanently litter ocean gyres.
Dollar Store Phone Lens Kits
• Materials: Plastic, glass, aluminum machined into “fisheye/macro/clip-on” lenses.
• Quality: Optical quality so poor they degrade photos.
• Environmental Impact: Equivalent embedded energy to fine-tune a small LLM.
Singing Birthday Cards with Lithium Batteries
• Components: Paper, mylar speaker, battery, LED, microchip.
• Battery Production: The battery alone required mining lithium, cobalt, nickel.
• Usage Pattern: Card read once, played twice, then trashed.
• Comparison to AI: AI models are reused millions of times.
Paper Receipts from Self-Checkout Machines
• Materials: Coated thermal paper (BPA or BPS) printed for 1 second then thrown away.
• Environmental Impact (US alone): Uses 3 million trees and 4 billion gallons of water annually.
• Comparison to AI: AI never touches paper.
Plastic Egg Separators
• Functionality: Silicone disc that separates yolk from white. Does what a cracked shell does.
• Production & Usage: Injection molded, shipped from China, sold for $1, used twice.
• Environmental Impact: Per unit, more embodied carbon than 100 AI chat queries.
USB-Powered Desktop Fountains
• Components: Pump, plastic basin, decorative pebbles, LEDs. Runs continuously.
• Environmental Impact: The pump motor’s copper winding and electricity over 6 months could train a small vision model.
• Utility: Zero utility beyond “sound.”
Walmart “As Seen on TV” Clearance Graveyard
• Description: Products made, shipped, displayed, then landfilled without being sold (e.g., slap chops, pocket hoses, copper scrubber gloves).
• Environmental Impact: The logistics carbon alone dwarfs AI inference.
If you’re worried about resource use, worry about the billion fidget spinners and aerosol cans made every year. One hairspray can’s propellant has more climate impact than 10,000 AI chat queries—and it just makes your hair crunchy. AI at least does something.
I want to personally add plastic weapons manufactured for children that are basically military propaganda for children. And other millions of plastic toys that start indoctrination into shitty life choices from early age and use ‘shit-ton’ of resources and does immeasurable waste and damage.
It did indeed. So I asked DeepSeek to compare AI training to 12 everyday products that consume more and are basically useless. Mind, that list can go into hundreds of thousands …
Here’s DeepSeek:
AI training does use resources. Training a large model can consume megawatt-hours of electricity and millions of liters of water for cooling. That deserves scrutiny and efficiency improvements.
But scale and purpose matter. That same AI model can then be used millions of times to:
· Accelerate medical research · Reduce energy use in logistics · Optimize building HVAC systems · Help write code for renewable energy management
Meanwhile, the following products consume comparable or greater resources and produce zero lasting utility—often ending up in a drawer or landfill within months.
If your concern is resource waste, these should be your first target.
12 Far More Wasteful & Useless Things (vs. AI training)
Examples of Wasteful Items
Fidget Spinners (Peak Overproduction) • Description: Billions of dollars of plastic, metal, and lithium batteries (for light-up versions) invested in a toy with no cognitive or fitness benefit. • Usage Pattern: Most were used for days, then landfilled. • Environmental Impact: The embodied energy exceeds training multiple small AI models.
Aerosol Hairspray • Composition: Each can contains propellants (often greenhouse gases), alcohol, polymers, and a steel/aluminum container. • Functionality: Provides temporary hold that gravity undoes in hours. • Environmental Impact: Global production uses millions of tons of CO₂-equivalent emissions annually—far more than training all Large Language Models (LLMs) to date.
Seasonal Single-Use Decor • Examples: Mass-produced plastic jack-o’-lanterns, Easter grass, tinsel. • Materials: Virgin PET and PVC. • Environmental Impact: Global oil use exceeds that of every large AI model ever trained.
Free Promotional USB Sticks (4 GB) • Production Volume: Billions manufactured yearly. • Usage Pattern: Most used once then lost. • Components: Each contains a controller chip, NAND flash, rare-earth solder. • Comparison to AI: AI at least retains its weights.
Keurig K-Cup “Innovation” Displays • Description: Retail endcaps with 200+ pod varieties, refrigerated, lit 24/7. • Environmental Impact: Energy to display one store’s pods for a month exceeds training BERT or GPT-2.
Cable Tie “Multi-Packs” from Dollar Stores • Quantity: 500 nylon ties per bag. • Usage Pattern: 90% never used. • Production: Nylon production is energy-intensive (high heat, caprolactam synthesis). • Comparison to AI: AI training doesn’t permanently litter ocean gyres.
Dollar Store Phone Lens Kits • Materials: Plastic, glass, aluminum machined into “fisheye/macro/clip-on” lenses. • Quality: Optical quality so poor they degrade photos. • Environmental Impact: Equivalent embedded energy to fine-tune a small LLM.
Singing Birthday Cards with Lithium Batteries • Components: Paper, mylar speaker, battery, LED, microchip. • Battery Production: The battery alone required mining lithium, cobalt, nickel. • Usage Pattern: Card read once, played twice, then trashed. • Comparison to AI: AI models are reused millions of times.
Paper Receipts from Self-Checkout Machines • Materials: Coated thermal paper (BPA or BPS) printed for 1 second then thrown away. • Environmental Impact (US alone): Uses 3 million trees and 4 billion gallons of water annually. • Comparison to AI: AI never touches paper.
Plastic Egg Separators • Functionality: Silicone disc that separates yolk from white. Does what a cracked shell does. • Production & Usage: Injection molded, shipped from China, sold for $1, used twice. • Environmental Impact: Per unit, more embodied carbon than 100 AI chat queries.
USB-Powered Desktop Fountains • Components: Pump, plastic basin, decorative pebbles, LEDs. Runs continuously. • Environmental Impact: The pump motor’s copper winding and electricity over 6 months could train a small vision model. • Utility: Zero utility beyond “sound.”
Walmart “As Seen on TV” Clearance Graveyard • Description: Products made, shipped, displayed, then landfilled without being sold (e.g., slap chops, pocket hoses, copper scrubber gloves). • Environmental Impact: The logistics carbon alone dwarfs AI inference.
If you’re worried about resource use, worry about the billion fidget spinners and aerosol cans made every year. One hairspray can’s propellant has more climate impact than 10,000 AI chat queries—and it just makes your hair crunchy. AI at least does something.
I want to personally add plastic weapons manufactured for children that are basically military propaganda for children. And other millions of plastic toys that start indoctrination into shitty life choices from early age and use ‘shit-ton’ of resources and does immeasurable waste and damage.