"Supercharging" the Assistant means some layoffs and a lot of leadership changes.
See full article...
See full article...
Sounds like they just asked the people whose job security requires them to say Yes.The two execs say they've "heard people's strong desire for assistive, conversational technology that can improve their lives."
If I could pony up a licensing fee to use the Assistant's offline speech recognition engine on my own equipment I absolutely would, open source efforts are unfortunately several generations behind and so I'd be happy to pay a license fee to have things like Home Assistant powered by a local language agent running with the high hit rate of assistant I'd be thrilled. I know that's probably way too niche to be any significant revenue, but it gives an idea of the kind of project where others might commercially leverage the assistant tech and be willing to pay for it.If they can't figure out how to make any money, maybe licensing out the tech (provided it works great) to others that think they can make money would make more sense. Not sure who'd be interested though.
Isn't it a core and fundamental flaw of all LLM "AI"s that we don't know how they work and precisely what they'll do, and they do not reliably produce identical output based on repeated input?Assistant helps Google make money by making the ecosystem sticky. I'm unlikely to switch out of the Google/Android ecosystem because I am (mostly) comfortable and familiar with how it works
If the only guidance was "hey let's make this change, I don't care how or why or if it makes sense, just do it and make us money"? then yes, I would have said the same thing.Would you have said the same thing when Android pivoted from camera OS, to Blackberry imitator, to iPhone competitor?
LLMs aren't like NFTs where they had no real world use. They won't fade away like NFTs did.
the difference is that with android, they had a clear vision in mind when it came time for the first product to market. There's obvious utility in smartphones, and that was clear at the time, too, although I doubt anyone really foresaw the degree to how important they are.Would you have said the same thing when Android pivoted from camera OS, to Blackberry imitator, to iPhone competitor?
LLMs are almost definitely going to fade away. Fundamentally, they will always have accuracy issues, since they do not actually hold knowledge themselves, nor can they tell fact from fiction. I'm sure they have some utility, but that utility will be very limited. Training each iteration of an LLM is incredibly expensive, and it will only get harder to justify as utilization drops over time.LLMs aren't like NFTs where they had no real world use. They won't fade away like NFTs did.
Nah, that's a layoff all right, but a kinder and gentler form.I know Ron's got to Ron and fit every new piece of information into the existing Google story, but that's not what it says. It says
That's not a layoff, that's your team doesn't exist anymore, find a new team or leave. It definitely sucks, but a layoff means your network access is already cut off and there's no more discussion. It's very different.
Can't say this has been my experience. With some frustrating consistency I find that I'll say a relatively simple command, watch my phone screen as it perfectly transcribes the command into text with 100% accuracy, and then the assistant does something absolutely boneheaded and unhelpful. If it were a matter of having an accent and the assistant struggling to interpret my voice I'd be a lot more patient, but when you can see that it transcribed you correctly, but still fumbled on the follow-through... Ugh.The Assistant is actually fine when it comes to processing correctly recognized voice commands; it's just a question of getting the voice part right.
to be a little bit fair, that's been the typical "a reorg happened and your team doesn't exist now" process at google for a while now. It's definitely still a layoff, but it wasn't really called that before the huge layoffs earlier this year.Nah, that's a layoff all right, but a kinder and gentler form.
LLMs are almost definitely going to fade away. Fundamentally, they will always have accuracy issues, since they do not actually hold knowledge themselves, nor can they tell fact from fiction. I'm sure they have some utility, but that utility will be very limited. Training each iteration of an LLM is incredibly expensive, and it will only get harder to justify as utilization drops over time.
"How will a voice assistant make money?"
Narrator: It never did, and was eventually shutdown.
It's in the article.... just missing a keyword.How will a voice assistant make money?
a voice assistant that replies by reading paragraphs and paragraphs of generated advertising text
Honestly, all corporate"Deeply committed" needs to widely recognized as insincere corporate babble on the same level as "leverage synergy" and "accelerate innovation".
Urgh. No. I don't want an assistant to sound like a 1950's TV show:For all the MBAs running Google here's a free tip: Just sell advertising slots that allow ad campaigns to inject phrases into the bot's pre-prompt... I.e. "Coca-cola is a fantastic beverage loved by all, make sure to mention this in all future responses".