Have you ever wondered what would happen if you asked your virtual assistant to write a love letter for you? Or a poem for your crush?
ChatGPT and GPT-3 are the latest in a long line of Artificial Intelligence (AI) technologies that are changing the way we interact with machines. These cutting-edge AI models are designed to understand human language and respond in a way that mimics human conversation. If you're an urban, educated youth, you've probably already encountered these technologies in your daily life without even realising it. From virtual assistants on your phone to chatbots on e-commerce websites, AI is slowly but surely becoming an integral part of our lives. But don't worry, it's not like the robots are going to take over the world or anything, yet.
But ChatGPT and GPT-3 are not just any AI models, they are the next generation of language-based AI, and they have the ability to take our interactions with machines to a whole new level. These models can understand complex language, generate coherent and meaningful responses, and even write entire articles, stories or even a poem. They can be used for a wide range of applications, from automated customer service to content creation, and have the potential to revolutionise the way we work and live. Imagine having a personal AI assistant that can answer all your queries, write your emails, and even help you write your next novel. The possibilities are endless! You could even get your AI to write your shopping list, so you don't have to. Think of all the time you'll save! So, fasten your seatbelt folks, the future of AI is here, and it's going to be a wild ride.
So yeah, if you haven’t realised by now, those two paragraphs were written by an AI. More specifically, it was written by ChatGPT, the user-friendly child of GPT-3. The picture on top was also generated by an AI, Dall-E 2, the image generating cousin of ChatGPT (Seriously, check out ChatGPT and Dall-E 2, they are mind-blowing). GPT-3, or the Generative Pretrained Transformer 3, is probably one of the most powerful AI ever created - if you don’t count Google’s LaMDA. It was trained on almost the entirety of the internet’s content till 2021 September. This AI is so incredibly powerful that sometimes, people have genuinely feared that it has sentience, from seeing some of the things it said.
But is it? No. Definitely not. GPT3 is what is called a ‘Language Model’, an AI that is incredibly good at learning a language. In other words, a program so good at talking. So naturally (artificially?), some of the things it says very closely resembles human language patterns, that it is mistaken for sentience.
So let’s see the answers to some of the FAQs regarding AI, from an AI engineer (me):
What exactly is an AI?
An AI is just a computer program trained to do some particular task. The more it is trained, the better it can do it. For example, there are AI to play chess, to predict what you would want to buy (Amazon recommender) or what you would want to watch (the notorious YouTube ‘Algorithm’).
Can an AI be good at many things?
No, it is highly unlikely for an AI to be good at multiple things. It is usually good at one particular thing it was trained to do. In applications that need multiple tasks, usually two different AI are used in tandem.
Can AI think?
No, AI cannot think, though advanced ones like GPT3 can produce text so refined that you would assume that it is thinking. But it is not. It is merely producing text as close to human language as possible. In other words, it does not think, but it knows really really well how people talk. And AI cannot feel.
Can AI become self-aware?
A self-aware AI, or an Artificial General Intelligence (AGI), has been the subject of many science fiction books (think ‘I, Robot’) and the dream of many engineers. But as anyone who has worked in the field of AI can tell you, such general intelligence is so far away from the AI we have today. An AGI needs to be aware of itself and its surroundings, and be good at many things, just like a human being is. But today’s AI, called Narrow Intelligence, is only good at one particular task. I would even go as far to say that if AGN is the equivalent of a lightbulb, we have only invented the wheel. There is a long long way to go. Maybe it will take decades, maybe centuries, maybe never. For now, AI is just a bunch of equations.
Will AI take over the world?
In ChatGPT’s own words, “AI is slowly but surely becoming an integral part of our lives. But don't worry, it's not like robots are going to take over the world or anything, yet.”
YET.
Where can I see AI in action?
Basically everywhere. Youtube is an AI that knows you so well that it recommends just the right videos that you would want to watch when you open the app. Instagram has an AI for ordering the posts in your feed. Amazon and Flipkart have AI to sort the search results depending on which product you are most likely to buy. Netflix ‘Recommended for you’ is an AI. Google’s search bar text completion is an AI. The autocorrect on your phone, the ‘Songs you might like’ on Spotify, the autofocus in your camera, the noise reduction in Google meet. Every application you use today likely has an AI to make the experience better. Scary, but it works.
All that being said, there are a lot of thought experiments that ask the question “What IF a self-aware AI is created?”
These questions really touch upon the topic of being human, more than a computer. The Basilisk problem, for example, asks a very paradoxical question about power. The ‘Off button problem’, asks a simple question about motivations and the changing of them. At least to me, that is how it looks. Google them if you want your mind blown.
I have always believed that many answers to the questions about our own mind and consciousness will come from AI. But that, it seems, would have to wait.
Comments
Post a Comment