Epistemic status: been thinking about this a contradiction I've seen in some of my friends views on AI and think its likely that a lot of people think in this way, especially in communities that really don't like it (eg. most of Lemmy). I know a ton of math and read/work a lot on AI theory+mech interp+safety, so I think I have a pretty strong understanding of the underlying mechanics. I'm fairly confident on the general ideas in this post, but as I explain at the end: a very nuanced understanding involves holding many of these otherwise contradictory viewpoints as true. Writing this because I genuinely want those against AI to have an easier time convincing people.
(for this AI=LLM)
It is not reasonable to think both that
- AIs are going to be forever incompetent or
- AIs cannot create new things (excluding creative works) or
- AIs cannot do very simple tasks
and
- AI will replace my job or
- AI will bring mass surveillance or
- AI can produce deepfakes that are 100% convincing or
- AI can be superpersuasive, manipulative (can cause AI psychosis)
(a lot of popular articles on AI fall into one of the above categories, and are often targeted towards the same people)
On another note, I have yet to hear a person against datacenters to give me a good reply to the question "Isn't the water just recycled back into the system?", and in fact they're usually uncertain why the datacenters need the water in the first place. I'm not some accelerationist and I don't think we need to cover the globe in datacenters, but those strongly against AI should have more concrete views than AI = bad, and then be able to back them up without having to do a search for sources, just from memory.
And yes there are exceptions to this in which holding both views make sense but this requires a lot more nuance but unless you are interested in this, that extra info probably would just clutter up your worldview.
I am not an expert but I know more about llms than the average end user. Let's go through everything.
AIs are going to be forever incompetent
Ai already isn't incompetent in proper use cases and is improving. HOWEVER when llms are used in inappropriate ways such as far customer service, or legal advice, which they frequently are, they are and will most likely remain harmful and incompetent. They're great for finding and predicting patterns and processing large amounts of data. The problem is that the people in charge of llm companies are forcing them in everything possible, even when they clearly have no reason to be used for certain things, which causes a ton of issuesm
AIs cannot create new things
This is a gray area. It technically creates new things but you need to remember that llms are not humans, what an llm is doing is compiling what it has in its database and regurgitating it back up to be used. When an llm "creates" an image what you're looking at is bits and pieces of its stored data. Llms do not have the capacity to create anything on their own without this base data. If an llm didn't have it's training, it would be a paper weight. This is where this argument comes from.
AIs cannot do very simple tasks
Entirely incorrect, it can and does many things. As stated before, it's great at processing large amount of data. It's extremely useful for researchers. People have found new planets with llm technology. However most end users only see llms when they are forced to interact with them, and in many of those situations, the llm is being used for something completely inappropriate for an llm, which does make it come off as completely useless. If the people responsible for pushing and deploying llns were not doing this, and were deploying them responsibly and ethically, I'm quite confident there would be far less hate for them.
AI will replace my job
Without knowing what your job is I can't tell you that but they can and have displaced many people's jobs, and has been a massive bane especially in the art and tech industry
AI will bring mass surveillance
Ai is already used frequently for mass surveillance.
AI can produce deepfakes that are 100% convincing
It's already alarmingly close to being able to do this
AI can be superpersuasive, manipulative
There are many cases of it doing this. Ai psychosis would not be a term if it didn't happen frequently enough to need a term
Llms can both be helpful and incredibly destructive. When it's used properly it can be a major boon, and when it's used incorrectly, or maliciously, it can and has destroyed lives.
Those datacenters are also extremely harmful to the communities they are built in. They make people's utility bills spike dramatically, cause immense harm to the environment, and yes, they do use up and pollute a massive amount of freshwater. They are also extremely loud and can cause sounds in frequencies we can't detect but can and do cause people to get ill. I encourage you to look up everything I've said here to confirm for yourself.
Anything can be both helpful and harmful. Uranium can be refined into fuel for very clean and efficient energy.... It can also be used to make bombs that are powerful enough to end life on earth. A hammer is great for driving nails into wood, but it can also be a murder weapon. Something can be both helpful and extremely harmful at the same time.
Thank you for coming to my Ted talk.
thanks for writing this
yeah this is what I meant by nuance, I just dislike when people have mostly-contradictory viewpoints that overlap in the sole fact that they believe 'AI=bad'. for example I've heard the same people irl both say things that show they believe 'AIs cannot do very simple tasks' and 'AI can be superpersuasive, manipulative'
Those things are not contradictory they just required nuance. If you put an llm in a customer assistance role, it will be almost entirely useless. One had a full on breakdown when it tried to run a vending machine in a test. But give an llm obscene amounts of information to sift through to look for patterns and it will thrive.
People tend to say llm is useless because people running the llm companies tend to push it into deployment in places it has absolutely no business being in, it's their only real experience with them, and it is usually forced on them without their consent making the interaction negative before it ever began.
It's not useless but many people encounter it in areas where it's either not helpful, or an active hindrance. They're working with the information and first hand experience they have and when they encounter it, they're usually correct in that circumstance that it is indeed useless. Nuance goes in both directions.