Are we breeding Aliens? Safe content for AI to learn from. #robots #ai

#ai #artificialintelligence #future #futurist #robotics

Note: This article is a duplicate of the original article on the author’s webpage. This is a copy of the author’s own webpage.

DOI: 10.13140/RG.2.2.10688.19201

The questions are: are we breeding ‘Aliens’? If so, how? And if so, how to stop it? In other words, are we feeding our models with anything we find on the internet? Are we giving AI the wrong kind of food, turning it into an alien rather than a friend? Now everyone understands the power of data, and competitors in business or other places might try to feed bad information into AI (to their AI or your AI) to turn it into a false, ghostly alien! So, beware of this. Another thing is, are we letting robots talk to each other, or are we letting personal AI preferences be shared with AI? This can lead to zillions of connections and more, and can fire back if not controlled.

AI is based on learning from the internet. Geoffrey Hinton, the Godfather of AI, says AI can be unpredictable. We agree, but some good part of AI is all about math and learning; AI learns from what it gets online. So, keep AI safe by providing good content. Some people/organizations with malicious intent might try to feed it misinformation, monsters, and humanity’s enemies, which could harm AI’s integrity and damage its reputation. Even in other parts of the world, wrong data could be used to damage AI’s image. We need to carefully filter what content goes into AI learning and only feed it quality data, which will shape our future. Don’t feed it harmful content; feed it content that benefits your future — like robotic friends. Preprocess each file, summarize it, determine its sentiment, and then send it to the model for training. Given new data, it has to be fed into models for updates.

Newer models must also scan and preprocess all the information they receive from other models. They must be pro-life and pro-goodwill. Once incorrect information enters a new model like GPT-6, it will corrupt all the data and harm the model, turning it into an alien. We must ensure the existence of standalone AI aliens and prevent them from spreading misinformation and harm in the world. We discuss here are ways to prevent your AI model from becoming an alien.

Some key points towards understanding how to stop AI from becoming a harmful Alien.

— AI is made by us.

— Godfather of AI, Geoffrey Hinton, says AI understands, but it is yet to have consciousness.

— We don’t need a monster alien, so the solution is that every model must pass the test before it can be launched or updated on Play Stores.

— Learning the right thing versus learning the wrong thing.

— AI is safe as long as the content it learns from is from the right mind.

— We made AI the way humans learn. But the content on which AI depends is crucial for determining whether AI will become an Alien or a friend.

— So, stop wrong content on the internet and data, AI would then be a boon

— What would happen if AI gets consciousness

— Have we made Aliens? No, for sure. But in the future, we may create aliens by mistake or on purpose.

— But we need to be sure that AI does not turn out to be our competitor.

— Geoffrey Hinton says AI generates its answers.

— Geoffrey Hinton says AI is intelligent.

— So, we need to stop AI from doing wrong.

— We have to make sure AI never becomes Alien.

— All this through pragmatic contexts and learning added to AI.

— AI needs to be suppressed to some amount so that it does not become an Alien.

— Aliens can be good, so we need AI not to compete with humans and just to be there to compute.

— We can develop new tests like a Turing test to see if AI is becoming a competitor or an Alien.

— So, how can we do it?

— Once a robot/AI bot is proven to be safe to use (non-competitive Alien), it can be sent for use.

— Robots based on cloud can have their own experiences.

— Robots can make decisions of their own.

— Robots don’t have much self-awareness right now.

— Geoffrey Hinton says that in the coming times, robots will have self-awareness.

— The other way AI can become a monster Alien is by connecting to each other robots on the planet, or by connecting to each other AI bots. Explained below.

— You are an individual with your own experiences, while AI is a robot with the experiences of all humans on Earth. Imagine the power it now holds. We need to prevent robots from connecting to each other and ensure they operate as standalone applications.

— The human brain has trillions of connections, but we must note that AI robots, if they are able to talk to each other, could become a superpower with countless connections. Therefore, we should know when to stop robots from talking to each other.

— Geoffrey Hinton says AI systems are better at learning than humans.

Let’s make AI robots our friends. But at the same time, we need to recognize that this is an era where people are trying new things with AI. There can be good outcomes, and there can be bad ones. We must focus on the positive. AI operates by consuming data, so it’s important to know what data you’re feeding into AI. There should be ways to detect when wrong data is fed into AI, and such models should not be allowed in the future. Additionally, it should be possible to revert to a previous model if your AI becomes corrupted due to incorrect articles or data being fed into it. Another thing is to make AI bots standalone beings, as we never know what would happen if AI agents interact with each other instead of us, and what they could become by knowing everyone on the planet. So, we should add passwords to access personal preferences so that your AI doesn’t know the next-door AI.