• 0 Posts
  • 28 Comments
Joined 1 year ago
cake
Cake day: June 5th, 2023

help-circle




  • Advertising is like the Kudzu vine: neat and potentially useful if maintained responsibly, but beyond capable of growing out of control and strangling the very landscape if you don’t constantly keep it in check. I think, for instance, that a podcast or over-the-air show running an ad-read with an affiliate link is fine for the most part, as long as it’s relatively unobtrusive and doesn’t put limitations on what the content would otherwise go over.

    The problem is that there needs to be a reset of advertiser expectations. Right now, they expect the return on investment that comes from hyper-specific and invasive data, and I don’t think you can get that same level of effectiveness without it. The current advertising model is entrenched, and the parasitic roots have eroded the foundation. Those roots will always be parasitic because that’s the nature of advertising, and the profit motive in general when unchecked.




  • I’d also consider myself pretty tech-savvy, but that came from plenty of mistakes growing up including putting malware on the family computer at least twice (mostly ads for these “Pokemon MMOs” back in the mid aughts that were too enticing for my kid brain to refuse 😅).

    It’s very easy for me to forget how much of an outlier my tech experience is among most folks around my age. I had an acquaintance in the first year of college I helped by giving essay advice, and was very surprised to see that the only thing they really knew how to do was basic use of apps on their iPhone. They got a laptop for school, but no computer experience, no keyboard typing experience, and even just the iPhone Settings app was a scary place to be avoided for the most part. To this person, Microsoft Word was a new thing they had to learn on top of everything else. In college. It was also in the South so I don’t know if I should be that surprised unfortunately.

    Regardless, it was pretty wild to me, but a very real reminder that not everyone has access to the same resources education, and/or experience to draw on.




  • The device wouldn’t necessarily have to be constantly streaming the audio to a central server. If it’s capable of hearing wake up words like “Ok Google” it’s capable of listening for other phrases and having onboard processing to relay back the results much more compressed. Whether or not this is common practice is another matter, and yes the algorithms are scary good even without eavesdropping.





  • That’s fair. I think fundamentally a false positive/negative isn’t that much different. Pretty much all tests—especially those dealing with real world conditions—are heuristic, as are all LLMs by necessity of the design. Hallucination is a pretty specific term given to AI as an attempt to assign agency to a system that doesn’t actually have any (by implying it’s crazy and making stuff up instead of a black box with deterministic inputs and outputs spitting out something factually wrong but with a similar format to what is trained on). I feel like the nature of any tool where “you can’t trust this to be entirely accurate” should have an umbrella term that encompasses both types of providing inaccurate info under certain conditions.

    I suppose the difference is that AI is a lot more likely to randomly go off, whereas a blood test is likelier to provide repeated false positives for the same person with their unique biology? There’s also the fact that most medical tests represent a true/false dichotomy or lookup table, whereas an LLM is given the entire bounds of language.

    Would an AI clustering algorithm (say, K-means for instance) giving an inaccurate diagnosis be a false positive/negative or a hallucination? These models can be programmed on a sliding scale and I feel like there’s definitely an area where the line could get pretty blurry.


  • I mean, AI is used in fraud detection pretty often; when it hits a false positive (which happens frequently on a population-level basis), is that not a hallucination of some sort? Obviously LLMs can go off the rails much further because it’s readable text, but any machine learning model will occasionally spit out really bad guesses almost any person could have done better with. (To be fair, humans are highly capable of really bad guesses too).