Search results

  1. B

    Man learns he’s being dumped via “dystopian” AI summary of texts

    Actually the AI might have done him a favor in this case. Somehow I have a feeling that the summary was less hurtful than the actual text. Only an asshole would break up that way.
  2. B

    OpenAI unveils easy voice assistant creation at 2024 developer event

    Please tell me how you run a 405B parameter model locally on a mac? Even a four bit quant would consume 200+ gigs of memory.
  3. B

    iPhone 16 and 16 Pro review: Camera features headline a mature product

    60 Hz is really unacceptable for a phone at this price point imho. And to think that fluidity of the user interface was one of the original iPhones big draws.
  4. B

    OpenAI’s Murati shocks with sudden departure announcement

    I don’t believe thats the case, I think this is Murai wanting to move on. Despite her being touted as pushing the board to remove Altman, according to internal memos leaked all she did was provide the board with candid feedback about Altman when asked by a board member. The same feedback had...
  5. B

    NIST proposes barring some of the most nonsensical password rules

    It’s not only about sanitation; If your password even goes near a database they are doing it wrong. Anyone who doesn’t know hashes and salts has no business building login systems at all.
  6. B

    Due to AI fakes, the “deep doubt” era is here

    Not sure I agree. There will always be lots of people looking to deceive but are poor at it. When the quality and cost of the tools for deception improves it stands to reason that there will be more deception. Also "only fools get fooled, and I am not a fool" is what every fool says. Everyone...
  7. B

    Due to AI fakes, the “deep doubt” era is here

    Even more damaging than misinformation imho is the discrediting of real information that an ever decreasing signal to noise ration has. The "trust no one" mantra is a perfect encapsulation of this. It sounds simple, but you can't have society without trust.
  8. B

    Amazon kills remote working, tells workers to be in office 5 days a week

    Amazon has gotten fat and inefficient. Executive leadership blames remote work and middle management. Guess they don’t have mirrors on the top floor.
  9. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    People skim documents, its akin to a quick and dirty summary. It’s not a binary thing, most use cases fall on a scale of accurate vs fast. But totally agree on framing LLMs as a replacement to humans as incorrect. Real value comes from augmenting humans with orthogonal intelligence.
  10. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    Again, depends on your use case. If you have lots of time to spend and mistakes are costly, then yes humans are the way to go. If it needs to be quick, cheap and right most of the time then LLMs are way superior. Lets say for instance you need to float potential issues in a document. An LLM...
  11. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    Your point is only valid if you have man hours to spend and the budget that goes with that. If you give the human and the LLM each ten seconds to summarize a 300 page document? Or to do a sentiment analysis of 100000 posts about your brand?
  12. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    Let’s say process and compress large amounts of knowledge in an instant; summarize, categorize etc as an example. There are many, many things in development now as I’m sure you are aware of. Unfortunately many of these things seem hell bent on replicating human intelligence instead of finding a...
  13. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    And that would be the core of the issue; it’s very hard for humans not to correlate language with general intelligence. In many ways an image generator is as impressive as an LLM from a technical perspective, yet no one would claim that there is intelligence hiding behind the output. It’s...
  14. B

    OpenAI’s new “reasoning” AI models are here: o1-preview and o1-mini

    I think one of the biggest problems with anthropomorphizing LLMs is the fact that we judge them by things that are trivial to humans, but make them seem dumb (strawberry rs that have been toted to death). Clearly the value of a new tool is not in the stuff it cannot do, but in the things it can...
  15. B

    Found: 280 Android apps that use OCR to steal cryptocurrency credentials

    If your PC is compromised to the point where an attacker could extract recall logs, they can likely spy on your screen regardless. Judging the utility of something based on downside alone is not really a good metric. If that would be the case, everyone would do their banking physically at the...
  16. B

    New AI standards group wants to make data scraping opt-in

    Isn’t this going to kill open source models? My primary fear with AI is that a few tech giants are going to own it all, and everything is going to be on their profit driven terms. This type of legislation unfortunately looks like it will play to that scenario.
  17. B

    FTC urged to make smart devices say how long they will be supported

    Agreed. And the best part about this is that it creates free marketing and awareness of the program by exposing consumers to it on boxes. Secondly it also serves as an incentive for companies to go beyond simply complying with the lowest level mandated by a law.
  18. B

    Audi replaces its bestseller—here’s the next Q5 SUV

    Oh, it’s objectively a much better car today for sure (emotionally, that V8 sound though, but I digress…). I’m just saying the power/performance gap to the base model has never been so slim on paper. Perhaps upsells to the S should worry Audi more.
  19. B

    Audi replaces its bestseller—here’s the next Q5 SUV

    Had an S5 back in 08 that made 350-ish hp. That S engine hasn’t received a power bump in almost twenty years now.
  20. B

    Hobbyists discover how to insert custom fonts into AI-generated images

    Agreed. The way to be productive right now is to generate some first drafts and then edit it yourself. However, I’m sure the same arguments were made by people when design started being done on the computer; It’s so much easier to do this change with a marker, right here on the paper. The...