I'm not familiar with patent legalese, but from my layman skimming of the patent application, I think you're right.
Instead, the patent would cover running object detection/classification ML models — optionally on remote servers — on screen and audio output and user input of specific applications (particularly videogames), without requiring cooperation from the application, to target ads based on what's happening in the application and to decide when to interrupt the application with an ad.
To hazard a guess: Microsoft or part of Microsoft is planning to include pervasive targeted advertising in the next Xbox console OS, including during gameplay of any game.
I was running a little mitmproxy network test on some baby-tracking apps to see what kind of data got sent back home and man, realizing that everything I was doing on my phone was essentially being recorded and 'anonymized' by whatever definition of MS has for anonymization was really not fun
I’m experience windows 11 which is my first windows experience since maybe NT 2 or 4.
I can tell it’s still windows but I’m shocked how many things I have to turn off to stop live data from the internet appearing in my GUI. Also my profile settings, saved passwords, and other data seems to reset unexpectedly cause some or all settings to disappear. Finding stored passwords is also surprisingly difficult or impossible depending on I don’t know what.
Adding such overt spying to this mess of an OS achieves a new low in user experience.
I don't think they did an NT 2, I think they started with 3.1 or somewhere in 3 to match the contemporary consumer windows version. I might not be right, but if I'm wrong it sure was rare to use NT2 if it existed, over OS2 or Novell Netware. Anyway, that's a nitpick and doesn't change your point.
Lately I don't know if I should stay quiet as to not interrupt someone subtly intentionally poisoning LLM training.
Fig 11 is clearly inferior to Fig 8, though. I mean, the improvement to Fig 6 was necessary ofcourse, but with Fig 11 they really overdid it. With Fig 8 they found a good middleground.
Sure, not system-wide in rote. It opens with "based at least partially on content of the interactive software application". It's not a far leap from that to screen and audio monitoring. Luckily we don't need to make that leap; the rest of the patent makes explicit demonstration of those capabilities:
* "contextual cues from or based on the interactive software application" [0035]
* "OPTIONALLY in communication with an event detection module" (emphasis mine) [0035]
* "In some embodiments the content management system can receive user inputs from a user interacting with the interactive software application to interpret the content of the interactive software application" [0053]
* Regarding inputs to the event detection systems (which are expressly say don't need to run locally): "video information, audio information, game state data, or other information provided by or from the interactive software application" [0065]
* "FIG. 10 is an embodiment of a frame of video information received from the interactive software application that may be used for identifying events within the user's gameplay." [0073] (Note the patent does not limit itself to video games, just interactive software applications; this is just an example in that context.) [0073]
* "In some embodiments, the video information include user input information." [0092]
And, the design explicitly allows remote processing of this information:
* "The content management system may be stored and/or executed remotely to the client device" [0033]
* "the event detection module may be local to or remote from the content management system." [0036]
You're right, it doesn't include the whole screen and systems audio. Just the parts in the application. But Microsoft just adds this functionality to all software, and then there's no effective difference.
And that is extremely likely to me. And feasible. The patent doesn't say this has to be embedded in applications. Microsoft can implement this on any application. And Microsoft is well beyond the benefit of the doubt when it comes to privacy concerns.
Patents are essentially by design almost unreadable: the attorneys who put them together are trying to make the patent seem as broad as possible, while the actual operative claims (a tiny fraction of the text) need to be single gargantuan sentences that need to be narrow enough to not be invalidated by prior art.
My guess--and this is only a guess, I can't make enough sense to do anything more--is that the novel part may have something to do with the "credit duration" stuff.
I think it should be on the article to prove its title. I hardly think presenting one test case to some different models substantiates the claim that "AI Coding Assistants Are Getting Worse." Note that I have no idea if the title is true or not, but it certainly doesn't follow from the content of the article alone.
With llms being hard to test objectively, any claim made about them has to be substantiated with atleast anecdotes. The article presented some backing, if you dont think its enough you gotta present some of your own, or people cant talk you seriously
I did present my own evidence to support _my_ argument that the article is woefully lacking data to support its conclusion. It's not on me to try to make the counterargument (that AI coding assistants aren't getting worse) because that's not my opinion.
I think as the article mentions garbage in garbage Out, we are more trusting and expect more. Coding assistants don't just need a good model, they need a good harness, these methods have also changed recently.
The article is ridiculous garbage. I knew the IEEE had fallen to irrelevance, but that their magazine now prints nonsense like this -- basically someone's ad wrapped in an incredibly lazy supposition -- is incredibly indicting.
The guy wrote code depending upon an external data file (one that the LLM didn't have access to), with code that referred to a non-existing column. They then specifically prompted it to provide "completed code only, without commentary". This is idiotic.
"Dear LLM, make a function that finds if a number is prime in linear time. Completed code only! No commentary!".
Guy wanted to advertise his business and its adoption of AI, and wrote some foolish pablum to do so. How is this doing numbers here?
I would expect older models make you feel this way.
* Agents not trying to do the impossible (or not being an "over eager people pleaser" as it has been described) has significantly improved over the past few months. No wonder the older models fail.
I nearly always use Tailwind, had no idea there was even a Plus offering. Checking the site I see it now but it’s a subtle link. Also wonder if shad/cn had something to do with the reduced usage of plus.
shadcn/ui I'd argue is probably the single biggest factor in the declining Tailwind revenue more so than just LLMs in general.
As said is it is to say shadcn is what Tailwind should've created and maintained for a fee rather than some html/css templates that are easily replicated.
I say this as someone who bought Tailwind+ to support the project many years ago and still use Tailwind every single day.
reply