5 ms·

I really don't understand why Alexa hasn't gotten any better, or why FSD is still in beta after so many years of promises from Elon.

by Lendal 4y ago

fullshark 4y ago
I was thinking almost every B2C tech product has gotten worse or stagnated the last 5 years. Even if the underlying technology may have improved (e.g. perhaps YouTube/FB content is hosted more efficiently now) the end product for the user has basically not improved or gotten worse.
randcraw 4y ago
I think it's largely because speech recognition operates on only one sentence at a time. If you need to clarify a term said earlier or provide an explanatory phrase describing an earlier one, it can't connect the second utterance to the first. No context is retained. That's a surprisingly big limitation in terms of human conversation. Until NLP can create a short-term memory model to retain info on prior utterances, Alexa's comprehension will be limited to self-contained single phrases -- spells and incantations.
- jimbokun 4y ago
  Which is strange because that's exactly what very large language models like GPT-3 are good at. Keeping the context and generating plausible next sentences, paragraphs, etc. Wonder what's blocking that from being applied to systems like Alexa.
  - groby_b 4y ago
    Unpredictability. You (Amazon) don't want your precious little bot say random things that land you in the press. And oh, GPT-3 will. Point two is that models that can be goaded to a specific place. So you'll need to account for that too, or... oh look, press. Point 3 is that they're trained on the Internet at large, and you don't want that in your house :)
    - randcraw 4y ago
      And if it chooses the wrong prior topic (among several possibilities) the smartbot could behave maddeningly obtusely. I suspect even the best of these today will have little idea what part of the conversation is more relevant to the human's focus than any other part. It's just as likely to turn its attention to an irrelevant 'token' that came before as a relevant one and say something obvious or fatuous. It's a lot easier to converse gracefully than to add useful info to a conversation. This is essentially what has made past intelligent assistants so dysfunctional (like Microsoft's Clippy). They couldn't model cognitive concepts like 'relevant', 'interesting', or 'necessary' relative to the active topic.
- visarga 4y ago
  > Until NLP can create a short-term memory model Transformers can fit about 4000 tokens in the context window. That's a lot of dialogue rounds. The problem is these models are too expensive to offer at scale. We almost never get to see the latest AI in big corporation products. It is always AI from a couple generations ago that is cheaper to run. When we get the latest and greatest, like GPT-3 and DallE, then it's super expensive to use, not free.
mdorazio 4y ago
For FSD, Elon is a master hype man and absolutely awful at understanding the complexity of areas outside his expertise. FSD, the "Alien Dreadnought" factory failure, $30k Model 3, cargo to Mars in 2022, etc. Those of us in the trenches of the autonomous car space usually have much more realistic timelines (and hardware requirements).
- njarboe 4y ago
  Maybe true, but you never make the shots you don't take. Space-X has had a reusable orbital booster for almost a decade and no one else has built one yet. Tesla continues to ramp up production of the 3 and Y at 50% year over year. Next year they will likely produce over 2 million cars. The run rate at the end of this year will be about that. Once wait times for Model 3/Y orders are eliminated you might see the $30k model 3 (inflation adjusted) in a year or two. Tesla margins are large and cost reduction are continuous. It would be silly for them to price them that cheap right now and just give resellers all of that profit. "Doing the impossible behind schedule"
incomingpain 4y ago
I unplugged my alexa yesterday. It bleeped about a notification which I wasnt expecting. It wasn't a notification, it was to tell me about a sale they have on. I tried to tell it to stop and it wouldnt stop. Hit each button on the thing and it wouldnt shut up. So now its unplugged, probably not getting plugged back in
makeitdouble 4y ago
Alexa’s generic model is incredibly close to Google Search. You input a full phrase and the search is processed, trying to guess what you exactly meant. Alexa apps only get more latitude by allowing series of phrases. But looking at Google Search, we see that even with all the money research and devs in the world that model stays fundamentally limited and only properly works for the simplest tasks. Beyond that you’d need precise feedback on what Alexa is doing, and ability to micro adjust accordingly. Like when talking to someone and adjusting our sentences mid-speech as we realize the person in front of us isn’t properly understanding what we’re saying.
georgeecollins 4y ago
I would read Rodney Brooks for insight into this: https://www.cs.dartmouth.edu/~ccpalmer/teaching/cs55/Resources/Reading/The%20Big%20Problem%20With%20Self-Driving%20Cars%20Is%20People%20-%20IEEE%20Spectrum.pdf https://www.cs.dartmouth.edu/~ccpalmer/teaching/cs55/Resourc...
giantrobot 4y ago
Alexa is good enough to run ads and feed Amazon's sales profile of you. So for Amazon it's about as good as it needs to be. As long as they can get people to buy them they don't need to improve them much. FSD's feasibility has been overhyped by Musk for years. It's not an impossible technology but it's also not one you can just bang together with some shitty webcams hoping AI will somehow make everything right. Technology doesn't improve on its own. It's developed by people that have motivations and limited by engineering prowess and physics. Tesla's FSD is still hype because they're incentivized to get your check for FSD and hand you an IOU. Amazon's current offering is an advertising/surveillance vehicle that does those jobs so they're not incentivized to improve it.

I really don't understand why Alexa hasn't gotten any better, or why FSD is still in beta after so many years of promises from Elon.

by Lendal 4y ago

I really don't understand why Alexa hasn't gotten any better, or why FSD is still in beta after so many years of promises from Elon.

fullshark 4y ago

I was thinking almost every B2C tech product has gotten worse or stagnated the last 5 years. Even if the underlying technology may have improved (e.g. perhaps YouTube/FB content is hosted more efficiently now) the end product for the user has basically not improved or gotten worse.

randcraw 4y ago

I think it's largely because speech recognition operates on only one sentence at a time. If you need to clarify a term said earlier or provide an explanatory phrase describing an earlier one, it can't connect the second utterance to the first. No context is retained. That's a surprisingly big limitation in terms of human conversation. Until NLP can create a short-term memory model to retain info on prior utterances, Alexa's comprehension will be limited to self-contained single phrases -- spells and incantations.

jimbokun 4y ago
Which is strange because that's exactly what very large language models like GPT-3 are good at. Keeping the context and generating plausible next sentences, paragraphs, etc. Wonder what's blocking that from being applied to systems like Alexa.
- groby_b 4y ago
  Unpredictability. You (Amazon) don't want your precious little bot say random things that land you in the press. And oh, GPT-3 will. Point two is that models that can be goaded to a specific place. So you'll need to account for that too, or... oh look, press. Point 3 is that they're trained on the Internet at large, and you don't want that in your house :)
  - randcraw 4y ago
    And if it chooses the wrong prior topic (among several possibilities) the smartbot could behave maddeningly obtusely. I suspect even the best of these today will have little idea what part of the conversation is more relevant to the human's focus than any other part. It's just as likely to turn its attention to an irrelevant 'token' that came before as a relevant one and say something obvious or fatuous. It's a lot easier to converse gracefully than to add useful info to a conversation. This is essentially what has made past intelligent assistants so dysfunctional (like Microsoft's Clippy). They couldn't model cognitive concepts like 'relevant', 'interesting', or 'necessary' relative to the active topic.
visarga 4y ago
> Until NLP can create a short-term memory model Transformers can fit about 4000 tokens in the context window. That's a lot of dialogue rounds. The problem is these models are too expensive to offer at scale. We almost never get to see the latest AI in big corporation products. It is always AI from a couple generations ago that is cheaper to run. When we get the latest and greatest, like GPT-3 and DallE, then it's super expensive to use, not free.

mdorazio 4y ago

For FSD, Elon is a master hype man and absolutely awful at understanding the complexity of areas outside his expertise. FSD, the "Alien Dreadnought" factory failure, $30k Model 3, cargo to Mars in 2022, etc. Those of us in the trenches of the autonomous car space usually have much more realistic timelines (and hardware requirements).

njarboe 4y ago
Maybe true, but you never make the shots you don't take. Space-X has had a reusable orbital booster for almost a decade and no one else has built one yet. Tesla continues to ramp up production of the 3 and Y at 50% year over year. Next year they will likely produce over 2 million cars. The run rate at the end of this year will be about that. Once wait times for Model 3/Y orders are eliminated you might see the $30k model 3 (inflation adjusted) in a year or two. Tesla margins are large and cost reduction are continuous. It would be silly for them to price them that cheap right now and just give resellers all of that profit. "Doing the impossible behind schedule"

incomingpain 4y ago

I unplugged my alexa yesterday. It bleeped about a notification which I wasnt expecting. It wasn't a notification, it was to tell me about a sale they have on. I tried to tell it to stop and it wouldnt stop. Hit each button on the thing and it wouldnt shut up. So now its unplugged, probably not getting plugged back in

makeitdouble 4y ago

Alexa’s generic model is incredibly close to Google Search. You input a full phrase and the search is processed, trying to guess what you exactly meant. Alexa apps only get more latitude by allowing series of phrases. But looking at Google Search, we see that even with all the money research and devs in the world that model stays fundamentally limited and only properly works for the simplest tasks. Beyond that you’d need precise feedback on what Alexa is doing, and ability to micro adjust accordingly. Like when talking to someone and adjusting our sentences mid-speech as we realize the person in front of us isn’t properly understanding what we’re saying.

georgeecollins 4y ago

I would read Rodney Brooks for insight into this: https://www.cs.dartmouth.edu/~ccpalmer/teaching/cs55/Resources/Reading/The%20Big%20Problem%20With%20Self-Driving%20Cars%20Is%20People%20-%20IEEE%20Spectrum.pdf https://www.cs.dartmouth.edu/~ccpalmer/teaching/cs55/Resourc...

giantrobot 4y ago

Alexa is good enough to run ads and feed Amazon's sales profile of you. So for Amazon it's about as good as it needs to be. As long as they can get people to buy them they don't need to improve them much. FSD's feasibility has been overhyped by Musk for years. It's not an impossible technology but it's also not one you can just bang together with some shitty webcams hoping AI will somehow make everything right. Technology doesn't improve on its own. It's developed by people that have motivations and limited by engineering prowess and physics. Tesla's FSD is still hype because they're incentivized to get your check for FSD and hand you an IOU. Amazon's current offering is an advertising/surveillance vehicle that does those jobs so they're not incentivized to improve it.