3 core reasons why Artificial Intelligence is not ready for world domination.

Artificial Intelligence or Machine Intelligence is a reality that mankind is waking up to. Every day, papers are flooded with news reports of machines on the verge of taking over, robots driving people jobless, digital assistants like Amazon’s Alexa, Microsoft’s Cortana, Apple’s Siri and Google’s Now invading our homes.

Man has introduced smart technologies in education, engineering, food processing, defence, manufacturing and scores of others, but they are just that – smart in the sense of reducing human errors due to boredom, lack of focus or tiredness.

Steps like Data Democratization, IPv6 for the Internet of Things, better Privacy Control have all steps that hint to a future dominated by AI, but the core AI engine is a far dream.

It makes us all wonder – how soon will AI or Artificial Intelligence take over the world?

Narayan Murthy paints a positive picture.

Elon Musk has equated AI to summoning the demon

AI has a long way to go before even resembling human intelligence.

So where do we stand? Here’s my opinion. I feel we are far away from AI led World Domination due to the 3 factors mentioned below

If-then, If-then and more If-then.

The best algorithms devised by humans are still based on elementary programming 'if-then' logic. Click To Tweet

This logic is used multiple times probably recursively.

Take, for example, the ability of Google driverless cars. Given a lot of different, yet finite, scenarios, engineers have mapped all scenarios into probable situations that require drastic action, require steady but necessary action, those that require no action but continuous observation and events that can be ignored.

Everything that the car cameras observe is passed via an algorithm that maps the situation into one of these scenarios. Drastic action requiring scenarios only have one output action – the car comes to an absolute halt. So where is the artificial intelligence here?

The brilliant Google engineers have envisaged every possible scenario and have put it through a smart If (this event), then (do this) else (do that) logic repeatedly till all scenarios are caught and dealt with.

Trust you are getting the point I am trying to make…Where is the intelligence here?

Speed being mistaken for Intelligence

Chess playing engines or alpha go playing engines are hitting news headlines constantly on how they are yet to lose a game. Are we not mistaking their speed for intelligence.

A machine with loads of RAM and CPU at its disposal can analyse millions of moves ahead and plan backwards to select the moves that maximize its chance of winning. A human plays by experience and gut feeling.

Brute force method, when employed at high speeds, creates an illusion of intelligence. Click To Tweet

Brute force method, when employed at high speeds, creates an illusion of intelligence. Opening a lock with 3 number combination when given to a machine takes less time than humans.

The machine employs brute force method which means it will try by starting at 000, 001, 002,…,199,200,201,…997,998,999. However, the speed at which this will be tried out will be overwhelming for a human being.

A human being will on the other hand attack the problem by looking at the context. Whose lock is it? What is the person’s name? What is his birth date? Can these be clues to his 3 number combination?

Hence, the real test of machine intelligence will be to solve problems by examining one's experiences so far, studying the context and eliminating the least probable options. Click To Tweet

That for me will be the beginning of artificial intelligence.

Can our algorithms write new algorithms by themselves?

We are writing smarter and faster algorithms.

But are we writing algorithms that can write new algorithms by themselves? Click To Tweet

Even if the answer is yes, how robust are these algorithms? How are they being tested? Can they be relied on? Not yet.

This for me will be ultimate test that Artificial Intelligence will need to clear. It should be intelligent enough to recognize, a hitherto un-encountered situation, as a problem. It should be able to analyse the problem and devise an attack strategy to solve it. Finally, it needs to design, test and deploy the algorithm to tackle the problem at hand.

Tackling, these 3 core issues would be key for any Artificial Intelligence to take over the world. I would like to bet that we are decades away if not centuries from seeing a world dominated by Artificial Intelligence.

[yikes-mailchimp form=”3″ title=”1″ description=”1″]