Skip to content

Trump says US will start talks with China on TikTok deal this week

Technology
4 4 0
  • 113 Stimmen
    10 Beiträge
    15 Aufrufe
    S
    I admire your positivity. I do not share it though, because from what I have seen, because even if there are open weights, the one with the biggest datacenter will in the future hold the most intelligent and performance model. Very similar to how even if storage space is very cheap today, large companies are holding all the data anyway. AI will go the same way, and thus the megacorps will and in some extent already are owning not only our data, but our thoughts and the ability to modify them. I mean, sponsored prompt injection is just the first thought modifying thing, imagine Google search sponsored hits, but instead it's a hyperconvincing AI response that subtly nudges you to a certain brand or way of thinking. Absolutely terrifies me, especially with all the research Meta has done on how to manipulate people's mood and behaviour through which social media posts they are presented with
  • 64 Stimmen
    4 Beiträge
    4 Aufrufe
    U
    Weird headline. Is it the city making this recommendation, or the... Despite universal opposition by the dozens of residents present at the meeting, commissioners voted to recommend changes to the city’s zoning laws to allow data centers in areas zoned for light industrial use and to rezone a 700-acre property from agricultural to light industrial to accommodate the construction of a hyperscale data center.
  • Was ist ChatGPT?

    Technology technology
    1
    0 Stimmen
    1 Beiträge
    10 Aufrufe
    Niemand hat geantwortet
  • 61 Stimmen
    11 Beiträge
    24 Aufrufe
    K
    If you use LLMs like they should be, i.e. as autocomplete, they're helpful. Classic autocomplete can't see me type "import" and correctly guess that I want to import a file that I just created, but Copilot can. You shouldn't expect it to understand code, but it can type more quickly than you and plug the right things in more often than not.
  • 180 Stimmen
    13 Beiträge
    5 Aufrufe
    D
    There is a huge difference between an algorithm using real world data to produce a score a panel of experts use to make a determination and using a LLM to screen candidates. One has verifiable reproducible results that can be checked and debated the other does not. The final call does not matter if a computer program using an unknown and unreproducible algorithm screens you out before this. This is what we are facing. Pre-determined decisions that human beings are not being held accountable to. Is this happening right now? Yes it is, without a doubt. People are no longer making a lot of healthcare decisions determining insurance coverage. Computers that are not accountable are. You may have some ability to disagree but for how long? Soon there will be no way to reach a human about an insurance decision. This is already happening. People should be very anxious. Hearing United Healthcare has been forging DNRs and has been denying things like treatment for stroke for elders is disgusting. We have major issues that are not going away and we are blatantly ignoring them.
  • 272 Stimmen
    80 Beiträge
    116 Aufrufe
    S
    that sub seems to be fully brigaded by bots from marketing team of closed-ai and preplexity
  • AI model collapse is not what we paid for

    Technology technology
    20
    1
    84 Stimmen
    20 Beiträge
    33 Aufrufe
    A
    I share your frustration. I went nuts about this the other day. It was in the context of searching on a discord server, rather than Google, but it was so aggravating because of the how the "I know better than you" is everywhere nowadays in tech. The discord server was a reading group, and I was searching for discussion regarding a recent book they'd studied, by someone named "Copi". At first, I didn't use quotation marks, and I found my results were swamped with messages that included the word "copy". At this point I was fairly chill and just added quotation marks to my query to emphasise that it definitely was "Copi" I wanted. I still was swamped with messages with "copy", and it drove me mad because there is literally no way to say "fucking use the terms I give you and not the ones you think I want". The software example you give is a great example of when it would be real great to be able to have this ability. TL;DR: Solidarity in rage
  • 88 Stimmen
    26 Beiträge
    59 Aufrufe
    M
    I really can't stand this guy. What a slag.