Chris Mims on the future of Web search; Sierra on customer relations; Scott Alexander on a forecasting bot; Ethan Mollick says to skate to where the puck is going
Where LLMs will shine first is in the digital screen realm. Anything a computer geek can do for you, an LLM / tool will be able to do, soon-ish.
Deciding which of 4 similar family photos is best might remain tough, but I’m waiting for that level of AI help before I ago thru the thousands of mostly duplicate digi photos clogging up my backups.
Since my company’s not paying, only the less capable free version is what I slightly play with.
Better search results would be great, but I haven’t seen that yet—so I think there’s too much hype until they get that. Free speec to text transcripts are also on my watch list, waiting for free versions for 80+ minutes.
One pattern that seems to work very well is to use the LLMs to create little tools made of traditional software. Then you’ve captured a series of insights and intentions in a plain text prompts but also straightforward software.
"But regardless, if more efficient search becomes the main use case for large language models, then I will consider this a major disappointment."
I don't have a subscription to WSJ and can't read the rest of the article but how do we get from Mims being excited about a new search capability to disappointment if it's the best AI offers? Why can't you just be happy there is one more possible benefit?
Something I don't understand. It seems you are saying that if I click on a website, the provider gets some kind of reward but if the AI tool does it there is no reward. Am I missing what the reward is or is a person-click somehow different than a bot-click? or something else?
Websites can make money by running ads that need people to see them, or on collecting user data that could be sold on, or has value in marketing and further targeted adverts.
Perhaps segment your expertise a bit - because you've had several phases of your life - and create an agent that searches for communities (including AI communities) that are interested in one or another area of contribution from you - engaging the representative AI in discussions in that venue, bringing the 'real' you into the conversation if it becomes particularly active/interesting.
I found a different workaround when calling AT&T's tech support. I tell the answer-bot that if it doesn't put me through to a human, I'll recycle it for beer cans.
The answer 'bot they use doesn't know how to deal with that, so it connects me to a human.
If you try this, be careful not to extrapolate it to anything that might be interpreted as a true threat.
Oddly enough, this doesn't work at my health insurer.
> the only reason I’m calling is to talk to a human because I’ve already eliminated all other self service options in the app or on the website
In that case, you are not the typical user of customer service. I say this having observed intelligent but non-technical senior citizens trying to navigate apps or websites, and the less intelligent the more trouble they will tend to have. Unfortunately for you, there is no way for non-typical users to identify themselves as such and get to skip over the chatbots and the scripted humans to real competent specialists, except perhaps by paying way more for premium support.
Where LLMs will shine first is in the digital screen realm. Anything a computer geek can do for you, an LLM / tool will be able to do, soon-ish.
Deciding which of 4 similar family photos is best might remain tough, but I’m waiting for that level of AI help before I ago thru the thousands of mostly duplicate digi photos clogging up my backups.
Since my company’s not paying, only the less capable free version is what I slightly play with.
Better search results would be great, but I haven’t seen that yet—so I think there’s too much hype until they get that. Free speec to text transcripts are also on my watch list, waiting for free versions for 80+ minutes.
One pattern that seems to work very well is to use the LLMs to create little tools made of traditional software. Then you’ve captured a series of insights and intentions in a plain text prompts but also straightforward software.
"But regardless, if more efficient search becomes the main use case for large language models, then I will consider this a major disappointment."
I don't have a subscription to WSJ and can't read the rest of the article but how do we get from Mims being excited about a new search capability to disappointment if it's the best AI offers? Why can't you just be happy there is one more possible benefit?
Something I don't understand. It seems you are saying that if I click on a website, the provider gets some kind of reward but if the AI tool does it there is no reward. Am I missing what the reward is or is a person-click somehow different than a bot-click? or something else?
Websites can make money by running ads that need people to see them, or on collecting user data that could be sold on, or has value in marketing and further targeted adverts.
Maybe I should start to think about building a virtual version of myself that can go onto a video platform and answer questions from subscribers 24-7.
That's what Bard DeLong claims to want.
I thought Google changed from Bard to Gemini?
"building a virtual version of myself that can go onto a video platform and answer questions from subscribers 24-7."
And non-subscribers (i.e. future subscribers!) :)
Perhaps segment your expertise a bit - because you've had several phases of your life - and create an agent that searches for communities (including AI communities) that are interested in one or another area of contribution from you - engaging the representative AI in discussions in that venue, bringing the 'real' you into the conversation if it becomes particularly active/interesting.
I found a different workaround when calling AT&T's tech support. I tell the answer-bot that if it doesn't put me through to a human, I'll recycle it for beer cans.
The answer 'bot they use doesn't know how to deal with that, so it connects me to a human.
If you try this, be careful not to extrapolate it to anything that might be interpreted as a true threat.
Oddly enough, this doesn't work at my health insurer.
> the only reason I’m calling is to talk to a human because I’ve already eliminated all other self service options in the app or on the website
In that case, you are not the typical user of customer service. I say this having observed intelligent but non-technical senior citizens trying to navigate apps or websites, and the less intelligent the more trouble they will tend to have. Unfortunately for you, there is no way for non-typical users to identify themselves as such and get to skip over the chatbots and the scripted humans to real competent specialists, except perhaps by paying way more for premium support.