4 ms·

AI can now emulate human behaviors – soon it will be dangerously good

8 points by myinnerbanjo 8y ago

cirgue 8y ago
The term AI seriously muddies the waters of this conversation. None of these technologies have the slightest whiff of independent agency, autonomy, or the ability to learn/adapt without the considerable talent and prodigious effort of people or teams of people. No one thinks that Photoshop's more sophisticated tools are AI, but effectively these tools are Photoshop for video and audio. These are sophisticated tools to be sure, but ultimately the problem has nothing to do with AI: They're fundamentally problems with malicious human actors exploiting the broken epistemology of the internet, which has been happening for decades. AI is, however, a great way to get clicks/pageviews for non-insights.
- tabtab 8y ago
  Terms are ultimately defined by users of them, not some official or scientific body, for good or bad. I recommend people stop complaining about usage of the term. If somebody comes up with a clear distinction between "strong" AI and "weak" AI, then we can complain about "correct" usage with some wind at our back. Until then, the vocabulary issue is not worth focusing on. I suspect AI progress will be incremental such that there will be no definite distinction, only shades of whatever traits we use to define it. It will be more like a child slowly growing up rather than C3PO just walking out the lab door one of these days. The current crop of well-known AI uses mostly statistical pattern matching, and thus lacks "common sense" as we typically think of it. But eventually the pattern matching will be integrated into things like the CYC database of common-sense facts and start to show more common sense. However, it will probably make silly mistakes for a good while.
  - cirgue 8y ago
    As a user of the term AI, and a professional who works with “AI”, I think the term as used in popular discourse is (in any sense, weak or strong) a way to grab the attention of your audience as opposed to a term which transfers useful meaning from author to reader. I recommend people keep complaining, and loudly, about it, because nonsense marketing language and masturbatory buzzword chasing are bad for society, toxic to discourse, an active hinderance to progress, and a convenient way to distract from and conceal the real issue at hand here. They are, however, really good for people that want to sell nonsense products.
tabtab 8y ago
People stopped trusting photos when Photoshop (and clones) became commonplace. They will likewise stop trusting videos when enough known fakes are floating around.
- dragonwriter 8y ago
  > People stopped trusting photos when Photoshop (and clones) became commonplace. People stopped trusting photos that challenge preconceived bias then, but they still trust them when they reinforce such biases.
  - tabtab 8y ago
    That's going to happen regardless, and people have been doing that with written works since the first writing appeared. However, the technique used to rely more on supernatural than "fake logic". It was more like, "Deity X says you are bad, and thus must be smitten." But replace Deity X with Hannity or Bill Maher and you get mostly the same thing.

8 points by myinnerbanjo 8y ago

The term AI seriously muddies the waters of this conversation. None of these technologies have the slightest whiff of independent agency, autonomy, or the ability to learn/adapt without the considerable talent and prodigious effort of people or teams of people. No one thinks that Photoshop's more sophisticated tools are AI, but effectively these tools are Photoshop for video and audio. These are sophisticated tools to be sure, but ultimately the problem has nothing to do with AI: They're fundamentally problems with malicious human actors exploiting the broken epistemology of the internet, which has been happening for decades. AI is, however, a great way to get clicks/pageviews for non-insights.

tabtab 8y ago
Terms are ultimately defined by users of them, not some official or scientific body, for good or bad. I recommend people stop complaining about usage of the term. If somebody comes up with a clear distinction between "strong" AI and "weak" AI, then we can complain about "correct" usage with some wind at our back. Until then, the vocabulary issue is not worth focusing on. I suspect AI progress will be incremental such that there will be no definite distinction, only shades of whatever traits we use to define it. It will be more like a child slowly growing up rather than C3PO just walking out the lab door one of these days. The current crop of well-known AI uses mostly statistical pattern matching, and thus lacks "common sense" as we typically think of it. But eventually the pattern matching will be integrated into things like the CYC database of common-sense facts and start to show more common sense. However, it will probably make silly mistakes for a good while.
- cirgue 8y ago
  As a user of the term AI, and a professional who works with “AI”, I think the term as used in popular discourse is (in any sense, weak or strong) a way to grab the attention of your audience as opposed to a term which transfers useful meaning from author to reader. I recommend people keep complaining, and loudly, about it, because nonsense marketing language and masturbatory buzzword chasing are bad for society, toxic to discourse, an active hinderance to progress, and a convenient way to distract from and conceal the real issue at hand here. They are, however, really good for people that want to sell nonsense products.

tabtab 8y ago

People stopped trusting photos when Photoshop (and clones) became commonplace. They will likewise stop trusting videos when enough known fakes are floating around.

dragonwriter 8y ago
> People stopped trusting photos when Photoshop (and clones) became commonplace. People stopped trusting photos that challenge preconceived bias then, but they still trust them when they reinforce such biases.
- tabtab 8y ago
  That's going to happen regardless, and people have been doing that with written works since the first writing appeared. However, the technique used to rely more on supernatural than "fake logic". It was more like, "Deity X says you are bad, and thus must be smitten." But replace Deity X with Hannity or Bill Maher and you get mostly the same thing.