Not_mikeyNot all ai is bad, just most of it
I am totally looking forward to AI customer support. The current model of a person reading a scripted response is painful and fucking awful and only rarely leads to a good resolution. I would LOVE an AI support where I could just describe the problem and it gives me answers and it only asks relevant follow up questions. I can’t wait.
They’re already deployed and they’re less than helpful, because LLMs are bullshitting machines.
I already use llms to problem solve issues that I’m having and they’re typically better than me punching questions into Google. I admit that I’ve once had an llm hallucinate while it was trying to solve a problem for me, but the vast majority of the time it has been quite helpful. That’s been my experience at least. YMMV.
If you think I suck, I’m guessing you haven’t actually used telephone tech support in the past 10 years. That’s a version of hell.
If you think LLMs suck, I’m guessing you haven’t actually used telephone tech support in the past 10 years. That’s a version of hell I wish on very few people.
I’m specifically claiming that they’re bullshit machines. i.e. they’re generating synthetic text without context or understanding. My experience with search engines and telephone support is way better than what any LLM fed me.
I understand their limitations, but you’re overselling the negative. They’re fucking awesome for what they can do, but they have drawbacks that you must be aware of. Just as it’s lame to be an AI fanboi, it’s equally lame to be an AI luddite.
It’s funny tou bring up luddites, since they actually had the right idea about technology like LLMs. They were highly skilled textile workers who opposed the introducyion of dangerous medhanical looms that produced low quality goos, but were so easy to use so that a child could work them (because they wanted to employ children). They only got their bad name of backward anti-technology lunatics afterwards. But they were actually concerned for low quality technology being deployed to weaken worker’s rights, cheapen products and make bosses even richer. That’s actually the main issue I have with what’s happening with AI.
There’s a book by Brian Merchant called “Blood in the machine” on the topic, if you’re interested. He’s also on a bunch of podcasts, if you’re not the big reader.
I’m referring to “bullshit” in the way argued in this paper:
Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that these falsehoods, and the overall activity of large language models, is better understood as bullshit in the sense explored by Frankfurt (On Bullshit, Princeton, 2005): the models are in an important way indifferent to the truth of their outputs.
ChatGPT is bullshit - Ethics and Information Technology
Recently, there has been considerable interest in large language models: machine learning systems which produce human-like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that these falsehoods, and the overall activity of large language models, is better understood as bullshit in the sense explored by Frankfurt (On Bullshit, Princeton, 2005): the models are in an important way indifferent to the truth of their outputs. We distinguish two ways in which the models can be said to be bullshitters, and argue that they clearly meet at least one of these definitions. We further argue that describing AI misrepresentations as bullshit is both a more useful and more accurate way of predicting and discussing the behaviour of these systems.