Ask HN: What are your AI fears?

2 points by pbrw 3 years ago | 8 comments

Recent progress in AI gives me anxious feelings which I can't address. What fears do you have?

My feelings:

- it's kind of a "cheat code" which makes a lot of mental work pointless. Why expanding your cognitive abilities if AI can do most of mental tasks faster and better?

- human creativity isn't any special thing

- giving more power to people eventually leads to a disaster (e.g. industrial revolution and climate change)

- I can't explain how AI can be so powerful and it makes me anxious

- jealous because I didn't choose an ML path at uni

qsort 3 years ago |

> - giving more power to people eventually leads to a disaster (e.g. industrial revolution and climate change)

This sure is a take.

> - jealous because I didn't choose an ML path at uni

Unless by "ML path at uni" you mean "PhD in ML/AI at a good research university", undergraduate-level "ML" is a trash fire and borderline scam. The bog-standard computer science curriculum as it would have been taught in the 80s is still the gold standard IMHO.

Now, on to the question. My fear is that we hyped AI too much. When inevitably the reality doesn't match the sky-high expectation, I fear the general public will associate AI and computer science in general with crypto, witch doctors and other assorted trash, which it very obviously isn't.

I see a skizoid attitude that combines attributing to AI capabilities it doesn't have, and refusing to accept what AI can do very well. Currently, the first part dominates the conversation, but tables can easily turn unless we start being more realistic.

Or maybe I'm just stupid and chatbots truly are that revolutionary, but somehow I doubt it.

anenefan 3 years ago |

Most fears would be based on how much AI will end up running.

My own fears:

AI will be used to further complicate dynamic page response to the user ... where it starts to become like asking a person ... if it likes the user fine, if it works out the user was in some field or line of work the response might be modified to accommodate or further muddy the waters ... if it arrived at hating the user, well time to blow some smoke up the user's arse. Some search engines are annoying enough with their clever algorithms to thwart earnest users, so I can imagine such dynamic websites driven by AI going the same path. I'm already a fan of static content sites ... and I imagine in the future people will start returning to them or use a search engines that will list only static page results.

AI will be used to create better and more unique scam sites, rather than current ones that sort of look the same and use nearly the same wording because often english isn't their first language.

Why I'm not that worried:

It's a tool. The day I can submit an old image based old pdf, and have it duplicated into a digital text along with the identical font, possibly images enhanced into colour or finer detail or better shading to fix a photo scanning mess -- I'm be tickled and an avid user.

In regard to programming, AI is limited to the current data available, and that's assuming the given data is correct, not just relying on what the person tasked to document the various dependencies or API that might be used.

theGeatZhopa 3 years ago |

I'm not in fear of AI. It will get powerful, no question, but it's more of a opportunity than a threat.

We will still have empowered humans, who can switch it off.

Or, the fear, that the creativity will decline is somehow flawed. Just think of the situation where it has been sayed about google ("no one will learn the stuff to have it handy when needed"). Where has it brought us to..? Exactly to this very situation.

But, we have a lot more knowledge accessible. That counts.

Also, with the cad and modelling software in mind.. this empowered the designers to construct better airplanes in less time. So, the guy's using the software don't need to learn the physics behind the airplane anymore. It is better, they learn to use the software:)

And.. every Generation is seeing some kind of doom because of the young ones.. to stupid, do not learn (or, learn not the proper things). Socrates said that. He would be surprised by our success.. and the youth adapted to new stuff.

No fear at all.

BillyTheMage 3 years ago | |

> We will still have empowered humans, who can switch it off.

The people with power are the people who make me afraid of AI

ofalkaed 3 years ago |

People who are driven to create will not cease being driven to create because AI is about. Unless AI somehow causes teenagers to cease trying to impress others the act of human creation is not going to cease. We will almost certainly have to suffer through yet another update of Cyrano de Bergerac* and certainly some will deal with ill gotten gains through AI but the human act of creation is still safe, worst case is that the act of creation will become more important than the creation, live and improvised art will take the lead.

*I was going to ask ChatGPT to write me a modern metafictional update of Cyrano de Bergerac with AI as Cyrano but apparently ChatGPT is getting overrun right now and there was no room for me and my request. I did get a good laugh about it asking me to confirm that I am human, will that simple question be enough to stop AI from taking over the internet?

theGeatZhopa 3 years ago | |

Not everyone will use AI. Some needs and will produce the data needed:)

RGamma 3 years ago |

That runaway-wise AI doesn't happen fast enough to save us and this planet from ourselves (including from pathological intermediate AI scenarios)

BillyTheMage 3 years ago |

My worst fears are that it makes Capitalism even worse. Capitalism already sucks the life out of everything. Now you're telling me we invented something to literally replace our creativity?

I fear that Capitalists will use it to extract exponentially more wealth from us with increased efficiency. Mass layoffs will increase competition for jobs. Labor will be much more alienated than it already is.

I fear that the Capitalists and the state will use AI to gain a level of control over people that can't be undone. They just have to run a simulation and be told by ChatCOP that ol' MountainMan is 99% likely to be a political dissident or terrorist or whatever they decide to frame thinking for yourself as.

I fear the Capitalists and the state will use it to create effective lies on a level they haven't been able to before. There is no truth anymore. They can generate any evidence for anything they want, and then they'll have another AI figure out which combination of words will make people believe it.

I fear that humans will be reduced even further to nothing but a fungible commodity to the capitalists.