Great article in the guardian. I know this will cause some cognitive dissonance here, but it’s worth considering perspectives that are the opposite of your own.
One solution to the problem is - tax the AI. Bill Gates talked about this awhile ago.
It can be a tax which starts out small at first and raised as AI starts to become more and more pervasive.
As far as I know OpenAI as a business should already be paying, Federal Income Tax, State and Local Taxes, Sales Tax, Property Tax and Employment Taxes.
I also have to pay my own taxes and VAT when I use OpenAI’s services.
When we’re talking about any specific AI tax we’re talking about adding extra cost to OpenAI’s users that will go to X government. Suppressing the desire for user’s and developer’s to create better and more efficient AI systems.
It’s really hard to create any form of “universal tax” system that applies to the whole world, meaning if a country creates a specific AI tax, they’re just hindering their own citizens use of AI.
The second article mentioned “taxing robots”, not AI. My experience here is many moons ago, I was working in a frozen pizza factory. Boring, backbreaking labor, but it paid well for a high school kid. One day, they decided to automate parts of my job with robots! I couldn’t tell you how happy I was! I later went to college and never looked back, and crossed out “factory job” from the list. So taxing this improvement is non-sensical, unless you want everyone to be enslaved to back breaking labor for the rest of their lives.
Which then brings me to the first article. I will say nobody really knows how the AI will pan out. But practical “boots on the ground” experience here is that it does remove a LOT of drudgery, which in-turn makes you more productive, but the author seems to vehemently disagree with this. Maybe if they lean-in and actually use the technology, and then tell me it was a horrible “tainted” experience or whatever, I might take the author more seriously.
Sam Altman suggested in the recent video that we should monitor GPU usage.
However, hardware solutions will always be hackable.
A simpler and probably more effective approach is just having a very small tax on compute, with serious penalties for avoidance.
This will have two benefits: it will mostly satisfy Sam’s desire to monitor and if AI starts getting out of hand in terms of social disruption, they can just crank up the tax.
A tax like this seems somewhat morally justified, as these companies are benefiting from content created by society at large.
Some exemptions for research would be ideal, at least enough to ensure we don’t hamper important efforts like curing cancer, solving climate change, etc.