Gary Marcus

Rank 14 of 47
|
Score 83
5/16/2024, 6:46:31 AM
X
In reply to:
Tim Spalding 馃嚭馃嚘
@librarythingtim
路
612d
@TerencePlizga @GaryMarcus @cshirky One thing I disagree with Dr. Marcus on there. He's vocal about wanting a more symbolic AI and in pointing out the many ways AI is dangerous, hurts creators, etc. Okay. But much of what he decries in current AI could ALSO be true of a better, symoblic AI.
Tim Spalding 馃嚭馃嚘
@librarythingtim
路
612d
@TerencePlizga @GaryMarcus @cshirky That LLMs are sometimes useful is a very different question from whether they are *NET* useful to the world. I'm certainly worried they will do more harm than good overall. Not the first time there for a technology, alas.
Tim Spalding 馃嚭馃嚘
@librarythingtim
路
612d
@TerencePlizga @GaryMarcus @cshirky I'm a deontologist in ethics, but this isn't an ethical issue. I'm just saying that there's no single answer to whether LLMs are useful or not. It's about the use. Unfortunately, yes, idiots and financially-motivated operators are and will be using them in dangerous ways.
Terence Plizga | terenceplizga@bsky.social
@TerencePlizga
路
612d
@librarythingtim @GaryMarcus @cshirky So you鈥檙e applying principles of utilitarianism, right? That鈥檚 fine, but it鈥檚 not an approach that I currently find compelling. And I worry that if you find a tiny window of utility, some money guy in Silicon Valley will try to drive a tractor trailer through it. ;)
Tim Spalding 馃嚭馃嚘
@librarythingtim
路
612d
@TerencePlizga @GaryMarcus @cshirky But that's irrelevant to the conversation I at least am having, which is about usefulness. Usefulness is not a matter of absolutes, but an empirical question, and will vary considerably according to what you want to do and how much risk you can tolerate.
Tim Spalding 馃嚭馃嚘
@librarythingtim
路
612d
@TerencePlizga @GaryMarcus @cshirky I don't know your view, but I would go farther than many, including鈥攕o far as I remember from his recent book鈥擠r. Marcus. I don't think real intelligence, or at least consciousness, is possible for any computer. Not now, and not in 1,000 years.

The statement is part of a broader conversation on Twitter about the utility and ethical considerations of large language models (LLMs) and artificial intelligence (AI) in general. The specific reply suggests that enhancing AI with the ability to fact-check and reason could lead to more effective safeguards against potential harms.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement promotes the idea of improving AI to prevent harm, aligning with the principle of striving to do no harm with words and actions.聽[+2]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    By suggesting improvements in AI, the statement implicitly supports the use of technology to foster a safer environment, which could promote understanding and compassion towards the potential risks associated with AI.聽[+1]
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement engages constructively in the ongoing dialogue about AI, contributing positively without resorting to personal attacks or ad hominem arguments.聽[+1]
  4. Principle 6:
    I will use my influence for the betterment of society.
    The suggestion to enhance AI capabilities reflects an intention to use influence for societal betterment, particularly in the context of technology ethics and safety.聽[+2]