The idea of Artificial Intelligence (AI) has the potential to be a great boon for society, but I fear it may have far more concerns than benefits. AI is increasingly being used in decision-making processes, from job applications to financial investments, and this could mean jobs of lower-skilled workers are lost or replaced by somewhat mindless machines. This could cause an economic recession due to those workers not having money to spend in the market. AI can also be used as a tool for propaganda and misinformation, making it easier for those in power to manipulate public opinion on a large scale. The game theory aspect of AI is also concerning; if machines are able to learn how humans think and make decisions based on that data, they could use it against us in unexpected ways. Despite the potential gains from using AI, I would argue that its drawbacks outweigh any positives and caution should come with its implementation.
Concerns from Lectures:
1. AI is more dangerous than nukes
2. Ask AI chat interesting questions and many surprising results come up. Ask follow up questions and the results get strange. Unbounded contradictions and complex moral and ethical dilemmas arise. Example: At a funeral the food is exceptionally catered. AI suggests more people should die.
3. AI has a presence of authority and dictates policy because it's authoritative in it's nature.
4. AI can generate red flags, false flags, propaganda, red hearings, and falsehoods easily. AI news would be very confusing and more like activism than news.
5. A bad decision is more likely to come from AI because humans can't see the cons of a complex situation. Example: Vaccine or nuclear strategy
6. AI linguistically is more of a simulation than creation.
7. AI is a disruptive technology but doesn't account for how economies work after AI.
8. AI is a bigger threat to white-collar jobs than to blue-collar jobs.
9. AI was asked to find new ice cream flavors:
Pumpkin trash basket
Peanut butter slime
Strawberry cream disease
10. AI tries to do what humans want but creates bigger problems because it never understood the original problem.
11. AI fails in unforeseen situations. Example: a tidal wave crashing into a nuclear power station.
12. AI is comand and control and not creation and innovation.
13. AI can identify a person in a picture but it doesn't know what a human is.
14. AI can identify a person in a picture but has no intuition. Example: A person in a crosswalk that is drunk or on drugs
15. AI will follow a command but won't do what we want it to do, often.
16. AI was asked to get a person from point A to Point B. AI grew the person to the length of line AB and pushed it over.
17. AI was set a task of design a robot body to cross terrain on Mars. AI decided to make robot legs 10 stories tall.
18. AI was tasked to design a human to run fast. The AI tried to create a human running backwards, rolling, falling down and sideways.
19. Humans give AI the wrong problem to solve. AI misinterprets the problem. AI was asked to come up with new paint colors: poop, suffer, gray public, turdly
20. AI was asked to identify a fish. Then AI was asked how it identified the fish. AI said, "fingers." The pictures of the fish had people holding the fish.
21. AI crashed an automated car because it was trained to only identify trucks from the rear and not profile.
22. AI can't forsee consequences because it never understood human beings to begin with.
YOU ARE READING
AI is more Dangerous Than Nukes
Non-FictionThis book reveals the unpredictable and potentially devastating consequences of AI, and how it can be used as a tool for propaganda and manipulation by the media and government. AI is more dangerous than nukes, and this book will make you fear it.
