I genuinely find LLMs to be helpful with a wide variety of tasks. I have never once found an NFT to be useful.
Here’s a random little example: I took a photo of my bookcase, with about 200 books on it, and had my LLM make a spreadsheet of all the books with their title, author, date of publication, cover art image, and estimated price. I then used this spreadsheet to mass upload them to Facebook Marketplace in bulk. In about 20 minutes I had over 200 facebook ads posted for every one of my books, which resulted in getting far more money than if I made one ad to sell all the books in bulk; I only had to do a quick review of the spreadsheet to fix any glaring issues. I also had it use some marketing psychology to write attractive descriptions for the ads.
My reaction to these actually useful cases is generally the same: That does sound handy, a time saver. If GenAI were free I’d say it’s amazing.
The problem is the cost, mostly the power cost. It’s just… Not worth it for something like scanning books. It’s almost always just not going to be worth it.
Have you looked at local LLMs? You can download and run them off your own machine - no connecting to an external server. They aren’t any more power intensive than a lot of video games. I downloaded a DeepSeek model and I use ComfyUI to operate it locally
I cant wait until they bring back the walgreens fridge door. AI data center speculating on your groceries so that they don’t have to actually gauge value.
Uber immediately ratifies any outliers and updates their pricing to reflect the conceptual value lost. Meaning for every cheap service you can find, it will increase the value of the whole product without coming back down as the value is speculated higher than before. Investor’s delight!
I genuinely find LLMs to be helpful with a wide variety of tasks. I have never once found an NFT to be useful.
Here’s a random little example: I took a photo of my bookcase, with about 200 books on it, and had my LLM make a spreadsheet of all the books with their title, author, date of publication, cover art image, and estimated price. I then used this spreadsheet to mass upload them to Facebook Marketplace in bulk. In about 20 minutes I had over 200 facebook ads posted for every one of my books, which resulted in getting far more money than if I made one ad to sell all the books in bulk; I only had to do a quick review of the spreadsheet to fix any glaring issues. I also had it use some marketing psychology to write attractive descriptions for the ads.
My reaction to these actually useful cases is generally the same: That does sound handy, a time saver. If GenAI were free I’d say it’s amazing.
The problem is the cost, mostly the power cost. It’s just… Not worth it for something like scanning books. It’s almost always just not going to be worth it.
Have you looked at local LLMs? You can download and run them off your own machine - no connecting to an external server. They aren’t any more power intensive than a lot of video games. I downloaded a DeepSeek model and I use ComfyUI to operate it locally
The way it writes marketing copy is absolutely perfect. It was so formulaic to begin with.
I cant wait until they bring back the walgreens fridge door. AI data center speculating on your groceries so that they don’t have to actually gauge value.
Uber immediately ratifies any outliers and updates their pricing to reflect the conceptual value lost. Meaning for every cheap service you can find, it will increase the value of the whole product without coming back down as the value is speculated higher than before. Investor’s delight!
It’s gonna be zimbabwe on crack.