Artificial Intelligence - The new Frankenstein?

From AI being the domain of professors & science heads to now being a mainstream debate on live television. It has come a long way.

Right now, the hottest discussion is rather showing a grim future as pointed out by the famous tv series "Black Mirror". Facebook had to shut down its AI operations when the AI bots developed its own language of communication. Articles on Elon Musk, Mark Zuckerberg, AI have been regularly featuring in print & digital mediums.

But how?

A recent article in Quartz by David Gershgorn reveals a little-known fragement of history, an annual geek-fest focused on image recognition that may very well be "ground zero" for modern AI. The ImageNet contest started in 2009, chiefly due to the passion of Professor Fei-Fei Li (Currently working at Stanford and Google).

It’s where AI-researchers battled for the prestige of having the algorithm with the lowest error rate.

Surprisingly within just 7 years, The winning accuracy in classsifying objects rose to a staggering 97.3% from 71.8%, eventually even surpassing human abilities" The real inflection point came in 2012, when one team beat the field by applying a Deep Learning approach, which would eventually become the gold standard for designing intelligent, and ultimately autonomous, systems capable of understanding natural language, diagnosing cancer and driving on the open road. One of the AI researchers involved with the event at the time, Matthew Zeiler, told author Gershgorn that the “ImageNet 2012 event was definitely what triggered the big explosion of Artificial Intelligence” 

Nowadays from TV sets to Microwave to Housing solutions, AI practically is under utilization everywhere. Big corporations are leading it.

1. Walmart is developing facial recognition technology to detect frustrated or unhappy shoppers
2. Softbank’s Pepper robot is greeting customers at Pizza Huts in China
3. Netflix’s personalization algorithms taps the behavioral aspects of the consumer thereby accounting for 75+% of viewership due to great recommendations.
4. Goldman Sachs' stock trading team has gone from 600 (at it's peak) to only 2 today, with coders and algorithms replacing traders and their intuition.

Gartner says that over 20% of operations in corporations is going to be done through AI. It’s pretty clear that the 4th industrial revolution is more than academic. 

As we are increasingly less surprised by announcements that machines have, again, out-performed humans in domains of intelligence long considered exclusively human, we are settling into a new state of being in which machines co-exist within our workplaces and families, thinking and doing, learning and improving, as people do. Or so it seems.  

When Google's DeepMind platform, AlphaGo, out-competed the world's best human Go player, even the top AI researchers were stunned.

The government of China recently announced a new formal policy making AI a national economic priority, which some believe was prompted by this gaming milestone. However, what is rarely discussed is the level of effort and infrastructure needed to achieve this victory. The system trained on 30M+ permutations of the game over two years, drawing down the equivalent of 1M watts of processing power (the human brain uses only 20 watts). It was a superb demonstration, but when developing an enterprise AI strategy, what's possible must be reconciled with what’s pragmatic.

Businesses typically do not have limitless budgets or cycles to prove that algorithms can beat humans. Use-cases must be vetted and prioritized based on an expansive readiness assessment -

Is there sufficiently good data available to train the model?

Do you have the right computational infrastructure?

Keeping in mind, the computational power of AI and its supremacy over human brain. Is it necessary to have an ethical code and a centralized infrastructure for all machines run on AI?

Lastly, the haunting questions always remains that if Artificial intelligence has the wherewithal to make its own language, what is the probability that it wont try to up the ante as its well equipped in understanding human behavior.

What if we create a product whose capabilities we are completely unaware of?

When "2001: a space odyssey" released in 1968, no one expected it will be so close to reality.

Similarly, 21st century movies & TV shows such as Black Mirror, Robot and other dark-themed movies on AI which have an intriguing plot might just be have a small note below its Title, stating that its "based on reality" in near future