4 ms·

I work at a startup in the AI space. We've been working with it for over a year now and I'm of the mind that AI, in it's current state, really isn't a tool mea

I work at a startup in the AI space. We've been working with it for over a year now and I'm of the mind that AI, in it's current state, really isn't a tool meant for an end user to interact with. It feels like it's the most useful when it's transparent and you -- as an end user -- don't know that there's an LLM in the process. The best use cases seem to be those that don't require an end consumer to directly interact with an AI, but their journey through some process is assisted by an AI instead. The problem is that a lot of the marketing -- like these examples and Google's very misguided "Write me a letter" to an Olympian -- exists because companies keep trying to make "fetch" happen; the AI becomes the journey instead of being an assistant in that journey. Amazon's Rufus is a prime example. It's tucked away as a button that one has to activate explicitly. But it would almost be better if it could just clean up the search results when I do a search and a bunch of junk is returned.

empath75 2y ago
There's basically two places where I see it as being useful -- as part of an automation pipeline where you need some translation layer from natural language to structured data or vice versa, and when the user knows it's an LLM and _wants to interact with an LLM_. I never want to see LLM generated content as an end user when I haven't specifically asked for it.
- CharlieDigital 2y ago
  I agree with those. The use case our startup is working is one more that I think really works well: why are you showing me these results? Our product basically takes any search output and uses our internal knowledge graph for our clients to "explain" to the customer "why these results for this query/question?" We use the same knowledge graph to then extend our client's native search results with much more relevant results using the LLM to better match results to the semantic intent of the question. It's all quite nifty and transparent; the user wouldn't know that there's an AI involved without labeling it as AI generated.
candiddevmike 2y ago
I absolutely despise rufus. I used to be able to search questions and reviews instantaneously. Now I need to wait for rufus to poorly summarize everything, tap at least 2-3 times, and then get to see the search results. Talk about piss poor UX.
- lubujackson 2y ago
  Reminds me of Austin Powers where he has a TV in his car in the 70s to talk with his boss then wakes up in modern times (the 90s) and has shitty, fractured streaming video. "Skating where the puck is going" is a great strategy for businesses - not so much for end users.
  - vundercind 2y ago
    It's similar to how ~all phone calls sound worse than they used to before the phone system was digitized. For a while, you could only get audio quality and delay that bad by using a cell phone (actually, those went through a couple phases of differently-bad audio quality, as the tech "advanced")! Then they brought that joyful new experience to land line phones, so everyone could have it. It's cheaper, though. I assume. The main effect of which is that it's finally cheap enough to be economical to phone-spam our grandparents and scam them out of their savings... hm, yeah, never mind, it's all down-side.
- ryanbuening 2y ago
  You're not alone. I search questions and reviews all the time. Rufus is terribly slow and doesn't serve the same function as my specific keyword search I'm looking for.
exitb 2y ago
But it seems like Apple could make "fetch" happen in the past. Their ads were very aspirational, almost like you could become an Olympian if you just owned a Macintosh. These ads though, they're not about "you can do more". They're about "you can care less".
- consf 2y ago
  [dead]
- pininja 2y ago
  These ads feel like YouTube Short sketches. A little absurd, and a bit of a fantasy land. Would people really interact this way - word for word? I think being aspirational with AI can feel a bit in-humane. A lot of people seeing this experience a bit of cringe when the topic of AI is brought up. Maybe they’re worried about their job, that technology is too pervasive - whatever it is, they’re uncomfortable. It’s an easier pill to swallow if I can laugh at it. Apple seems to be picking up on that in these ads.
- snovymgodym 2y ago
  Eh, I mean Apple has had its flops over the years like any company. Their ads were good, but all of their successful products were of high quality when evaluated objectively, in addition to having been marketed well.
consf 2y ago
I think a good marketing move is showcasing the seamlessness of artificial intelligence. For some reason, AI is being stigmatized
- CharlieDigital 2y ago
  Yeah, a really basic and relevant example might be if I have a meeting coming up on my calendar, an agent can pull info from my LinkedIn about the participants, check their socials (so I can make small talk and chat), research info about some of the topics in the meeting (e.g. pull stats, recent articles, etc.) and summarize it to prepare me for the meeting. Like, that's actually useful but probably not super sexy.
  - Ensorceled 2y ago
    This is a good example of where we should have stigma attached to AI: a partner showed me their CRM that summarized my profile for them. It was creepy in how far into my online life it had dug. It also said, along with the creepy things, that I was "risk adverse" and "only acted when I had complete information" ... as a startup CTO this is both laughably wrong and could have career implications.
    - CharlieDigital 2y ago
      LLM's can be tuned to do what you want it to do. It can be as simple as "summarize these notes" to as questionable as "rate this candidate for this job". The former is mostly just a reduction in task and condensation of information. The latter is the opposite: an extrapolation of some conclusion based on given information. This is not the fault of an LLM, but the people using it and the use cases that have been designed. An AI assistant that just gave me the facts (condense) rather than generate its own conclusions (expand, extrapolate) is fine, IMO.
      - Ensorceled 2y ago
        I get what you are saying but, the same way ad targeting can be used to "serve me more interesting ads" and "surveillance" can be used to find missing children and sex trafficing victims, it isn't being used that way across the board.

empath75 2y ago

There's basically two places where I see it as being useful -- as part of an automation pipeline where you need some translation layer from natural language to structured data or vice versa, and when the user knows it's an LLM and _wants to interact with an LLM_. I never want to see LLM generated content as an end user when I haven't specifically asked for it.

CharlieDigital 2y ago
I agree with those. The use case our startup is working is one more that I think really works well: why are you showing me these results? Our product basically takes any search output and uses our internal knowledge graph for our clients to "explain" to the customer "why these results for this query/question?" We use the same knowledge graph to then extend our client's native search results with much more relevant results using the LLM to better match results to the semantic intent of the question. It's all quite nifty and transparent; the user wouldn't know that there's an AI involved without labeling it as AI generated.

candiddevmike 2y ago

I absolutely despise rufus. I used to be able to search questions and reviews instantaneously. Now I need to wait for rufus to poorly summarize everything, tap at least 2-3 times, and then get to see the search results. Talk about piss poor UX.

lubujackson 2y ago
Reminds me of Austin Powers where he has a TV in his car in the 70s to talk with his boss then wakes up in modern times (the 90s) and has shitty, fractured streaming video. "Skating where the puck is going" is a great strategy for businesses - not so much for end users.
- vundercind 2y ago
  It's similar to how ~all phone calls sound worse than they used to before the phone system was digitized. For a while, you could only get audio quality and delay that bad by using a cell phone (actually, those went through a couple phases of differently-bad audio quality, as the tech "advanced")! Then they brought that joyful new experience to land line phones, so everyone could have it. It's cheaper, though. I assume. The main effect of which is that it's finally cheap enough to be economical to phone-spam our grandparents and scam them out of their savings... hm, yeah, never mind, it's all down-side.
ryanbuening 2y ago
You're not alone. I search questions and reviews all the time. Rufus is terribly slow and doesn't serve the same function as my specific keyword search I'm looking for.

exitb 2y ago

But it seems like Apple could make "fetch" happen in the past. Their ads were very aspirational, almost like you could become an Olympian if you just owned a Macintosh. These ads though, they're not about "you can do more". They're about "you can care less".

consf 2y ago
[dead]
pininja 2y ago
These ads feel like YouTube Short sketches. A little absurd, and a bit of a fantasy land. Would people really interact this way - word for word? I think being aspirational with AI can feel a bit in-humane. A lot of people seeing this experience a bit of cringe when the topic of AI is brought up. Maybe they’re worried about their job, that technology is too pervasive - whatever it is, they’re uncomfortable. It’s an easier pill to swallow if I can laugh at it. Apple seems to be picking up on that in these ads.
snovymgodym 2y ago
Eh, I mean Apple has had its flops over the years like any company. Their ads were good, but all of their successful products were of high quality when evaluated objectively, in addition to having been marketed well.

consf 2y ago

I think a good marketing move is showcasing the seamlessness of artificial intelligence. For some reason, AI is being stigmatized

CharlieDigital 2y ago
Yeah, a really basic and relevant example might be if I have a meeting coming up on my calendar, an agent can pull info from my LinkedIn about the participants, check their socials (so I can make small talk and chat), research info about some of the topics in the meeting (e.g. pull stats, recent articles, etc.) and summarize it to prepare me for the meeting. Like, that's actually useful but probably not super sexy.
- Ensorceled 2y ago
  This is a good example of where we should have stigma attached to AI: a partner showed me their CRM that summarized my profile for them. It was creepy in how far into my online life it had dug. It also said, along with the creepy things, that I was "risk adverse" and "only acted when I had complete information" ... as a startup CTO this is both laughably wrong and could have career implications.
  - CharlieDigital 2y ago
    LLM's can be tuned to do what you want it to do. It can be as simple as "summarize these notes" to as questionable as "rate this candidate for this job". The former is mostly just a reduction in task and condensation of information. The latter is the opposite: an extrapolation of some conclusion based on given information. This is not the fault of an LLM, but the people using it and the use cases that have been designed. An AI assistant that just gave me the facts (condense) rather than generate its own conclusions (expand, extrapolate) is fine, IMO.
    - Ensorceled 2y ago
      I get what you are saying but, the same way ad targeting can be used to "serve me more interesting ads" and "surveillance" can be used to find missing children and sex trafficing victims, it isn't being used that way across the board.