If someone you barely know offers to pay for whatever car you feel like buying, you’re likely to spend their money buying yourself a really nice car and not spend much effort negotiating.
"Most people pay for AI access through a flat monthly subscription—$20, $100, $200 a month depending on the service. Ask Claude to redraft your contract seventeen times. Ask ChatGPT to make an image, make it bigger. No wait, make it smaller. Add a joke. Remove the joke. The price is the same. You feel no friction because you bear no marginal cost."
This isn't quite right. I have my own Claude $20/month plan. That's the minimum charge. Claude runs in 5 hour increments that begin the moment you submit a prompt. If your prompts use up too much compute during a 5-hour session, you have to wait until the next 5-hour increment starts. Or, you can pay more for the overage until the next 5 hour session starts.
True, if your employer is paying, the user doesn't bear the cost. But, if the employer doesn't want that to happen, it can disallow overages.
The point is that there are checks and balances, even on a basic monthly plan.
there are controls for moral hazard. Like insurance companies have different tiers of prescription drugs, or require your physician to get pre-approval for a procedure (which is like the physician serving as an unpaid member of the insurance companies cost containment program--there is more to that story by the way). Claude has usage limits for sure... And thanks for pointing that out. I didn't know that the low paid subscription cut off at 5 hours.. I thought that was the free one. But the key point (that you make) is that suppliers of AI are indeed putting in place controls.. They are watching out, a bit, for where moral hazard might cause them problems. But the corporate consumers of AI are going to catch on more slowly... They may blame the overuse on all sorts of things except moral hazard, which is exactly how people mistakenly view the medical system.
Agree. An accurate, articulate summation of where we are. And the root cause of it all.....drum roll...government; or rather massive, overwhelming, corrupt government.
Moral Hazard for sure is built into the system, the incentives are not aligned with the greater good and things get worst, ultimately for everyone, but last, much later, for the fat cats living off the hazard. Too true. Well written. Let's all open our eyes and push for correction and restraint.
very concise article Dr. Hunt - great comparison with the medical system...
Great article.
Loved it.
"Most people pay for AI access through a flat monthly subscription—$20, $100, $200 a month depending on the service. Ask Claude to redraft your contract seventeen times. Ask ChatGPT to make an image, make it bigger. No wait, make it smaller. Add a joke. Remove the joke. The price is the same. You feel no friction because you bear no marginal cost."
This isn't quite right. I have my own Claude $20/month plan. That's the minimum charge. Claude runs in 5 hour increments that begin the moment you submit a prompt. If your prompts use up too much compute during a 5-hour session, you have to wait until the next 5-hour increment starts. Or, you can pay more for the overage until the next 5 hour session starts.
True, if your employer is paying, the user doesn't bear the cost. But, if the employer doesn't want that to happen, it can disallow overages.
The point is that there are checks and balances, even on a basic monthly plan.
there are controls for moral hazard. Like insurance companies have different tiers of prescription drugs, or require your physician to get pre-approval for a procedure (which is like the physician serving as an unpaid member of the insurance companies cost containment program--there is more to that story by the way). Claude has usage limits for sure... And thanks for pointing that out. I didn't know that the low paid subscription cut off at 5 hours.. I thought that was the free one. But the key point (that you make) is that suppliers of AI are indeed putting in place controls.. They are watching out, a bit, for where moral hazard might cause them problems. But the corporate consumers of AI are going to catch on more slowly... They may blame the overuse on all sorts of things except moral hazard, which is exactly how people mistakenly view the medical system.
Extremely interesting. Thank you, John.
Agree. An accurate, articulate summation of where we are. And the root cause of it all.....drum roll...government; or rather massive, overwhelming, corrupt government.
And interspersed throughout it all: human nature.
So where do we start?
Moral Hazard for sure is built into the system, the incentives are not aligned with the greater good and things get worst, ultimately for everyone, but last, much later, for the fat cats living off the hazard. Too true. Well written. Let's all open our eyes and push for correction and restraint.