“Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last.” – Stephen Hawking
Development of full artificial intelligence could spell the end of the human race.
These were words of Stephen Hawking, sharing his experience with the BBC after the launch of a new software system designed to help him communicate more easily. People are worried about AI taking over humanity. And there are some who deny it altogether.
Artificial Intelligence. It’s a term that promises a future of limitless possibilities and unprecedented challenges. But what exactly is AI, and should we be worried about it?
### [Title Screen: “AI: Revolution or Catastrophe?”]
### [Segment 1: What is AI?]
[Interview Segment:Age of AI]
Narrator
In simple terms, AI refers to a machine’s ability to mimic human intelligence. But what is intelligence?
### [Segment 2: What is Intelligence?]
**Narrator:**
To understand AI, we first need to understand intelligence. Intelligence is the ability to learn, acquire knowledge, solve problems, and adapt to new situations.
### [Segment 3: Does AI Really Have Intelligence?]
Does AI truly possess intelligence, or is it simply following programmed instructions?
The way AI learns, acquires knowledge and experience is called Machine Learning. AI learns and acquires knowledge through different ways and these are broadly categorised into following three types:
-
Supervised learning
-
Unsupervised learning and
-
Reinforcement learning
These are not the only ways through which AI learns.
Supervised Learning: In this method AI is trained on labelled data: means input and output are predefined. For example, recognizing whether there is a cat in the image or not. The AI model will be shown thousands of pictures of cats which are labelled as cat or no cat. With time AI learns and gets better in recognising whether there is a cat in the image or not. Whenever I open my camera app on my phone it tells me what type of picture I am taking.[Video clip of your phone showing AI] This is an example of Supervised Learning.
Unsupervised Learning: In this method AI is trained on large amounts of data which is not labelled. The AI model clusters or groups the data based on similarities.
For example in my phone’s gallery app, my gallery has clustered or grouped my photos based on similar features. It grouped images based on faces, and other features. So all the pictures of a person whichever folder they are in are grouped together in one place in my gallery. Similarly there are groups of cats, food, sunrise, sunset, and flowers. This is unsupervised learning.
Reinforcement Learning: In this method AI is trained to interact with the environment and learn by making mistakes. In this method AI is not programmed for a certain task but it learns on its own by doing.
For example Here
Here AI is given a goal or task to accomplish, and letting it free in the environment to learn on its own.
Decision Making: One of the prime motives of building machines or AI is to make better decisions and AI is pretty good at it. For example web search. When we search for a keyword in the search engine, the AI decides what information it has to show.
Experience: There is no point in building AI if it can not learn from mistakes. Every time AI makes a decision, it stores information whether it worked or not. So it learns through experience, and does it very quickly by running hundreds of thousands of iterations of a task. For example AlphaZero learns chess by playing thousands of games and gets very good at it over time.
So we can correlate the intelligence of machines with that of humans.[Video of Human brain and Robots side by side] So in reality AI has intelligence similar to that of humans. Means they learn, apply their knowledge in solving problems and get better over time by learning from mistakes. [Video clip showing AI robot making mistakes]
### [Segment 4: Types of AI]
**Narrator:**
AI can be categorised into different types based on its capabilities:
-
Narrow or Weak AI (ANI): This represents AI that can perform very basic decision-making functions, such as smart speakers (Siri), self-driving cars, google assistant and web search.
[Generative AI: A subtype of narrow AI which is capable of creating content, like images, songs, videos, etc., that closely resembles human-generated content. Examples include ChatGPT, Gemini, Midjourney, and DALL-E.]
-
Strong or General AI (AGI): This represents AI that can do lots of tasks like humans.
We haven’t come close to achieving this level of AI yet, but there are questions about its consequences for us.
-
Super AI: Then there is Super AI that can surpass human capabilities. It’s currently a distant dream.
### [Segment 6: The Age of Generative AI]
**Narrator:**
We are currently in the age of Generative AI. Generative AI is a subtype of Narrow AI, which can do specific tasks with human level expertise or even better. Tools like OpenAI’s ChatGPT, Google’s Gemini, and Midjourney can create human-like text, images, and more. These technologies are not just impressive—they’re transformative.
[Video clips of AI, ChatGPT, and Midjourney etc.]
### [Segment 7: AGI, Super AI, and Sentience]
**Narrator:**
Looking to the future, AGI represents the ultimate goal of AI development. AGI would be as versatile as a human, while Super AI would surpass human intelligence. But can AI ever be truly sentient or conscious?*
Sentience and consciousness are complex and poorly understood even in humans. Whether AI can ever achieve these states remains an open question, with profound implications for our future.
[Interview with Kai Fu Lee]
### [Segment 8: Hallucination and the Black Box Phenomenon]
**Narrator:**
AI isn’t without its flaws. One notable issue is hallucination—where AI generates incorrect or biassed information, and presents it as fact. To check this flaw, I gave ChatGPT a prompt and it created a response that was perfect, but one thing, it was as if it happened in real but it was all its own making.
Another challenge is the ‘black box’ phenomenon, where AI’s decision-making processes are opaque and difficult to understand.
**[Interview Segment:t]**
There is an aspect in this which we call the black box……
### [Segment 9: The Impact on the Job Market]
The most intriguing and worrisome thing about AI is its ability to replace the human workforce. According to a research by Mckinsey Global Research Institute around 400 to 800 mn jobs are replaceable by automation by 2030.
What is the solution for it? We will have to have some kind of universal basic income.
That is necessary in some situations, but do you think you need a UBI, probably you shouldn’t aim for that rather make yourself resilient to automation. The way you can do that is to upskill yourself.
There are both free and paid courses available online through which you can upskill yourself. One of the best platforms is Coursera, on which you can get premium courses for free, and learn from industry experts directly.
[Video clip of Coursera website]
Another similar platform is edX.
There is also a silver lining : According to the same report by Mckinsey Global Institute Around 555 to 890 million jobs will be created by rising income by 2030.
Governments can also play a part in imparting education that is future oriented.
### [Closing Scene: Inspirational Music and Visuals of AI Achievements and Ethical Considerations]
**Narrator:**
Artificial Intelligence stands at the crossroads of potential and peril. Its future will be shaped by our choices and how we harness its power.
As we navigate this brave new world, one thing is certain: AI is not just a technological marvel; it is a reflection of our hopes, challenges, and aspirations for the future.
**[Text on Screen: “The Future is Now. Embrace the Intelligence.”]**
**Narrator:**
*This is the age of AI. Welcome to the future.*
### [End Credits Roll]
**[Text on Screen: “Produced by [Your Name/Company]”]**