It takes longer than a search query, but they don't want the hassle of coordinating logistics, and have some peace of mind that they're getting what they asked for.
- they take a cut
- they are neutral because they want repeat bussiness
- they are personalised around your requirements
- they take care of the little details
It's exactly what I'm describing here, except automated. Because at heart their work is analyst work and thats exactly what you can automate with a LLM
Because it would be the less profitable bussiness.
>Also, to this day, LLMs/ANNs are not capable of explaining their outcome, and you won't argue/reason with them
How would I be able to tell that your response is human and not a LLM?
I definately can argue within a chat thread. They can also provide a plausible sounding explanation.
We know due to the nature of the technology that that explanation (sounding rational, deductive) isn't the truth.
We are just not all ready to accept all the same shortcomings also apply to humans.
How is this different from Google?
> Because it would be the less profitable bussiness.
How so? Considering that almost everything on the web right now runs on ad revenue, why would they pass on the opportunity to cash on that if they do no worse than the rest? Especially considering how lucrative those referrals might be, with our lack of preparedness countering those with traditional ad blocking techniques and being more susceptible due to inexperience.
> We are just not all ready to accept all the same shortcomings also apply to humans.
Because they do not. If you hire a human to do a job for you, you are bound by a contract (and protections, and recourses and enforcement). In the end you will be able to know whether you got robbed into a referral scheme and that's bad for business. With this kind of AI, the result is all you get (technically there is no way to explain it), which gives unfair deniability. Regulators might end up ruling whole classes of algorithms as unfit for purpose/deceptive, and I would personally applaud that. We are just at the beginning of all that and we should focus on developing the technology so it keeps working for us reliably, and not against us or under undetermined circumstances.
The devil is always in the details. A travel agency builds contacts with airlines, hotels, tour groups etc, so they can have a backup option available if your first preference is not. That requires humans.
An AI system that is simply browsing through the same websites a human would is going to come up with a lot of "no" responses for specific booking requests, or will simply quote higher prices once the booking websites figure out how to detect them.
Perhaps they will try to up the price on this platform compared to others. But only until AirBnB comes to compete on this platform too (and uploads their API integration)
It's a platform game.
There was a time when Google tried to index everything and afterwards came a time when everything tried to be indexed by Google. Once you are the default, they have to be there and they will have to compete with others there.
And as long as the chat protects users from losing control over their intent, that's where the users will be and that's where you will need to be offering your products to be viable. Obviously at that point the regulators come knocking..
Assuming there would only be one player. There won't be, but for now only OpenAI is actually aware of the opportunity and Microsoft doesn't even seem to understand it's a platform game now and not an ad game.