It is a great research companion, but it is perhaps the most important use case I've run across to feed multiple models from different providers the same/similar prompts. You'll get wildly different suggestions sometimes.
> Over time, we expect users to be able to delegate purchases entirely to their agent by setting clear parameters—such as spending limits, purchase categories, or time periods—and pre-approve those purchases with enhanced safeguards
Researching with AI is one thing, but the announcement is suggesting that eventually they want to be able to take the human out of that loop too.
This is not a good combination. We, the tech community, maybe can tell, to good extent... general population, cannot.
This just makes it so you can actually buy the thing from chatgpt vs. going to a specific site and filling out your info
automatable commodity is referring to buying stuff
Does ChatGPT know that my kid wants grapes but my wife wants him to eat only organic grapes, but doesn't like grapes herself, so we should only buy a small quantity of organic grapes?
Or that as a household we try to avoid Nestle products altogether?
Or that we want tortillas, but only made of flour and the burrito size from one specific manufacturer?
Maybe after enough time and enough coaching it could get it right, but I also have never ordered groceries online. I don't even trust a random person to do a good job picking produce, so I suppose I'm far from the imagined target market here.
This is yet another attempt at squeezing the money out of sellers.. and it will mostly be smaller ones that will be impacted.
One silver-lining is this might give some negotiating power to sellers on Amazon.. who have been ripped-off by Amazon more and more every passing year.. until of course, they both don't end up colluding.
Not saying a user couldn't also be tricked by this, but of course it is more nuanced than the models sometimes make it out to be.
Just saying something has not worked in the past doesn't make it true for the next iteration too. Many touch-enabled phones failed before the iphone.
I think you've got it backwards. The vast majority of people who don't have enough time to begin with will gladly just keep typing their shopping cart into their AI of choice.
I wouldn't be surprised if this ends up being the majority of usage for suburban housewives, along with webmd type things and relationship questions.
It turns out that people like to do some research, look at a lot of images, read user reviews and more before making a purchasing decision. The subset of people that treat most everyday shopping as a completely automatable commodity is much lower than the tech world thinks.