Did You Know – There are 7 Types of AI?

Did you know that #artificialintelligence has types? Did you know that some of it may not have been put to practical use yet? Will 2023 see the birth of such AI?

AI is classified as per its #capabilities or as per its #functionalities.

Let’s deep dive: -

AI as per capabilities:

1.????Narrow AI: -

• Also called Weak AI
• Focuses on one task at a time
• Will find it challenging to address tasks which are beyond its capabilities
• Classifies data and finds patterns
• Examples include Translate functionality on #google , or recommendations to buy related products when you shop on #Amazon etc.

2.????General AI:

Also known as Strong AI

• Will come with human-like cognitive capabilities
• Will understand intent and would be capable of making smarter decisions
• Will need little or no human supervision (can run completely unsupervised)
• Uses clustering and association with humongous volume of data
• Will advance its knowledge with each interaction/ experience and enhance its skills over time
• General AI has received a $1 billion investment from #Microsoft through #OpenAI. Fujitsu’s K computer and Tianhe-2 developed by the China's National University of Defense Technology are attempts at building Strong AI but there’s still a long way to go before we can claim any significant progress.

I read this very interesting article by Bernard Marr about Practical Examples of Strong AI?

Is #ChatGPT is an example of General or Strong AI? Thoughts here.

3.????Super AI:

• It can surpass human intelligence and productivity
• It can act like a human – it has emotions, needs, desires, values, beliefs
• It would be able to think, use experiences, create a feasible solution, solve problems, judge, and decide in a logical and well-informed manner
• Have we created SuperAI? Not yet. But do you think a SuperAI (which we may have seen in sci-fi flicks) will come to life in 2023?

Let’s now look at AI classified as per functionalities

Based on their function, AI can be of four types:

1.????Reactive Machine

• As the name suggests, it works only with present data
• Does not used stored interactions or experiences to act
• They are suited for just a specific task and can’t go beyond that
• They will have a defined output based on an input
• An employee query bot, #Netflix recommendation engine etc. are examples of reactive machines
• It can easily automate a clearly defined rule-based, redundant, and repetitive task and ensure that time, effort, and bandwidth of critical resources such as employees can be freed up

?

2.????Limited Memory AI

• It can absorb, retain, and reuse learning data enabling it to improve with time and usage
• With limited Memory AI, the AI environment is made conducive for the models to get self-trained and updated as per the patterns and behaviour of the model
• It can use data to make well informed decisions
• #selfdrivingcars are a good example of limited memory AI. They are trained to rely on their intelligence to understand its operating environment and take decisions with respect to safe navigation of the vehicle

3.????Theory of Mind AI

• Theory of Mind AI enables the machine to decide efficiently like smart human beings
• It can read human expressions, convert it into an understanding of emotions, and react accordingly
• While theory of mind AI is yet in the development stage, a humanoid bot Sophia seems to have capability to see and comprehend emotions and give a suitable reaction (Read more about Sophia)
• A recent Indian Language (Malayalam) movie has a similar Theory of Mind AI type of character which is central to the plot. It captures the story of a conservative villager who hates the idea of his son relocating to Russia for a job. However, when his son brings a robot to care for him, he develops an unlikely bond with the machine. Read more about the movie here.

4.????Self-Aware AI

• Such AI is aware of the needs, emotions, context, and intent of people around them
• They also have the self-awareness to modulate their responses, reactions, and actions as per their own desires
• While we don’t have self-aware AI today, as technology advances and innovation becomes more cost-efficient, it could create super intelligent AI

Some questions: -

• Will humans and self-aware AI co-exist and collaborate? What will we do if such AI goes rogue?
• Can it be controlled by their software, or will it be capable of mutating itself?
• Will Self-aware AI pose challenges to us – its creators?
• If self-aware AI does something unlawful or illegal, who is to be blamed?

As AI developments advance, so will issues and debates around it. Will there be regulations around the use of AI or guidelines that define ethical use of the same? Let’s see how it unfolds…