& watching the VergeCast watch The Browser Company. & watching the VergeCast say: “Imminent death of the web. Film at 11”. But this time it is really, truly, for real. The Browser Company’s Dia...
I couldn't really grok Arc. I've switched to Orion, mainly because I like Kagi's search engine, and want to support them.
I also couldn't grok this discussion at all. I'm mainly interested in supporting the Web against apps, mainly for privacy reasons. For the moment, my limited use of AI has transformed my experience of the Web.
There is certainly a lot of industry groupthink about the future of the web and how we will use it. Browsers that are infused with AI chatbots. Perhaps AI decoupled from screens (despite 2 major failures)? AMZN is pushing "reorders" more insistently, both through the browser and its distributed Echo devices. It is also using AI in its fulfillment centers, and I read today that it is testing humanoid robots to increase deliveries from its Rivian vans.
I cannot believe that people will accept AIs as the next, more advanced algorithmic approach to choose what you see and how you interact in the world. This is very much a P K Dick dystopia. AI needs to be unobtrusive and a help when asked. Unfortunately, it seems that it will gravitate to being used for ever more intrusive surveillance in both the workplace and in public. [OTOH, AIs may be a good way to speak truth to power without exposing yourself.]
Do you remember Amazon reorder buttons? You pushed a button reorder e.g. laundry soap. I wonder when nostalgic collectors will start driving up their price on ebay.
Apple creates an AI bot that effectively combines and replaces both Safari and Google search. It is embedded in the I/O. Users can select whatever browser and search engine they want, but it doesn't matter if they typically use the Apple bot, thus end running both EU rules and wiping out a big chunk of Google's most profitable market share. Although it doesn't end run China's AI rule.
Imagine that the Apple bot isn't so filled with ads and contorted results that it is more useful, particularly if it doesn't hallucinate. Apple doesn't overstuff it with ads because of their sense of decorum, and also because they make money from the device and app sales. Not a prediction, I have no idea what Apple is up to.
Most search engines already have a chatbot or something awfully similar. Most browsers already have a combined input box for URLs and search queries which could produce search results and/or provide a chatbot answer. It sounds like they're offering a smoother ride by putting four wheels on their innovative new automobile.
There's no way that they are going to be able to provide a useful chatbot. It would violate too many business models or run into too many privacy risks. You can't ask it questions about your own finances or medical history. You can't ask it to perform particular actions on websites like a natural language programmed bot. You can't ask it to post a picture on social media without running into a felony violation of business model.
It seems the real alignment problem isn't about humans and superintelligent AGIs. It's about the humans trying to do stuff on the web and the humans trying to make money providing AI services. No wonder Apple is having so much trouble coming up with a way of integrating AI. Their biggest selling point is that your computer works for you while every AI company out there expects its AI to work for them.
There are ways of making AI useful, but it's not the AI provided by LLMs. It's old fashioned AI using ontological frames and inferencing engines. An LLM component might be useful, but, as Apple has been discovering, LLMs aren't as smart as Calendar, Reminders or Contacts. That's a pretty low bar.
I couldn't really grok Arc. I've switched to Orion, mainly because I like Kagi's search engine, and want to support them.
I also couldn't grok this discussion at all. I'm mainly interested in supporting the Web against apps, mainly for privacy reasons. For the moment, my limited use of AI has transformed my experience of the Web.
There is certainly a lot of industry groupthink about the future of the web and how we will use it. Browsers that are infused with AI chatbots. Perhaps AI decoupled from screens (despite 2 major failures)? AMZN is pushing "reorders" more insistently, both through the browser and its distributed Echo devices. It is also using AI in its fulfillment centers, and I read today that it is testing humanoid robots to increase deliveries from its Rivian vans.
I cannot believe that people will accept AIs as the next, more advanced algorithmic approach to choose what you see and how you interact in the world. This is very much a P K Dick dystopia. AI needs to be unobtrusive and a help when asked. Unfortunately, it seems that it will gravitate to being used for ever more intrusive surveillance in both the workplace and in public. [OTOH, AIs may be a good way to speak truth to power without exposing yourself.]
Do you remember Amazon reorder buttons? You pushed a button reorder e.g. laundry soap. I wonder when nostalgic collectors will start driving up their price on ebay.
Apple creates an AI bot that effectively combines and replaces both Safari and Google search. It is embedded in the I/O. Users can select whatever browser and search engine they want, but it doesn't matter if they typically use the Apple bot, thus end running both EU rules and wiping out a big chunk of Google's most profitable market share. Although it doesn't end run China's AI rule.
Imagine that the Apple bot isn't so filled with ads and contorted results that it is more useful, particularly if it doesn't hallucinate. Apple doesn't overstuff it with ads because of their sense of decorum, and also because they make money from the device and app sales. Not a prediction, I have no idea what Apple is up to.
None of this seems to make any sense.
Most search engines already have a chatbot or something awfully similar. Most browsers already have a combined input box for URLs and search queries which could produce search results and/or provide a chatbot answer. It sounds like they're offering a smoother ride by putting four wheels on their innovative new automobile.
There's no way that they are going to be able to provide a useful chatbot. It would violate too many business models or run into too many privacy risks. You can't ask it questions about your own finances or medical history. You can't ask it to perform particular actions on websites like a natural language programmed bot. You can't ask it to post a picture on social media without running into a felony violation of business model.
It seems the real alignment problem isn't about humans and superintelligent AGIs. It's about the humans trying to do stuff on the web and the humans trying to make money providing AI services. No wonder Apple is having so much trouble coming up with a way of integrating AI. Their biggest selling point is that your computer works for you while every AI company out there expects its AI to work for them.
There are ways of making AI useful, but it's not the AI provided by LLMs. It's old fashioned AI using ontological frames and inferencing engines. An LLM component might be useful, but, as Apple has been discovering, LLMs aren't as smart as Calendar, Reminders or Contacts. That's a pretty low bar.