And deep learning, machine learning, cognitive and more…

Artificial intelligence (AI) is one of the hottest topics on Tech St. right now – and why wouldn’t it be given the exciting advances being made? But in our conversations about AI, we’re getting ahead of ourselves and mixing up progress toward the goal with having reached that goal. I write this piece to share some thoughts with you about where we’re trying to get to with AI vs. where we’re at right now. I hope you’ll jump into the conversation!

Let’s start off our discussion about AI by defining what it means (according to the dictionary and commonly agreed upon definition):
“The theory and development of computer systems able to perform tasks that normally require human intelligence, such as (1) visual perception, (2) speech recognition, (3) decision-making, and (4) translation between languages.”

Next, let’s take a closer look at three of the elements of this definition:

• Visual perception: an innate understanding of incoming visual information
• Speech recognition: not only being a speech front end, but also understanding speech
• Decision-making: based on incoming information, the ability to autonomously and intelligently plan next steps

Now, compare this to current technologies that are being called or misinterpreted as “AI.” Today, most often we have devices and systems that have voice recognition front ends and can complete tasks on command. But this is not AI. Take Siri as an example. Siri executes via voice commands what you otherwise would have typed. You and I would agree that Siri does not have any of the three characteristics stated above, right? We have sophisticated programs that can complete tasks, but ultimately these are pre-programmed algorithms, not AI. AI is not available as a consumer product yet. So we need to be mindful when we use the term AI.

Today when I walk into my home and ask Alexa via Echo (voice front end) to carry out tasks, my verbal commands immediately get converted into programs that carry out the tasks. The more connected the components, the more tasks the programs can complete. So I can ask Alexa to look through my music library, find pieces by Beethoven which are piano pieces, and maybe add another layer of depth or two. In fact Echo is running a program. Today’s Alexa is not seeing me, recognizing my mood, deciding what I want to listen to, nor does it independently carry out the tasks. Which of the above 3 elements does Alexa succeed at? Any?

Ultimately, in AI we want to enable the machine to think and behave as the human brain. A human brain takes millions of pieces of diverse input and determines what step to take next, while understanding the consequences of many parallel possibilities. That is very different than what we have today. A human brain can imagine and has context and relevance. All of which has to be built in a true AI system.

You have probably also heard the phrase “We are in another AI winter.” Anything disruptive goes through a period of disbelief and doubt. Sometimes it does not emerge out, but most often it does. I don’t think we are in an AI Winter; rather, we are trying to figure out how to deploy AI meaningfully. As an example think about 3D Printing. This technology started about 4 decades ago, and just became popular. Here is another example: AR/VR has been around for years, it is now highly visible. Often these types of technologies are predicated by funding and market visibility.

At the same time, be very cautions of marketing hype around AI, Cognitive and Machine Learning. Always explore beneath the covers to see what is real and what is just hype.

I am extremely optimistic about where we are headed. Big Data, data science, machine and deep learning are all moving in the right direction. However, I get worried when I see things like voice front-end technologies being misrepresented as AI. We still lack true AI for consumers. Even some of our most advanced autonomous technologies, like self-driving cars, are running highly complex back end processes and algorithms (After all, we still don’t have an AI in the car that can see an unknown sign and surmise what it is – a task that a human can do automatically with human intelligence and instincts) and not still truly autonomous.

Over a decade ago, we saw Big Blue beat Kasparov in chess. (Was Big Blue thinking independently?) Then we saw Watson beat Jeopardy players – but was Watson thinking independently with its own brain processing or was everything pre-programmed as algorithms? Only recently in AlphaGo’s triumph against the world’s leading Go player did we see highly complex and perhaps autonomous (not programmatic) thinking. 2016 was a WOW moment for that alone! Indeed amazing, but we have a long way left to go.

Yes, we are now finally in the world of deep learning, machine learning and algorithms that will learn, evolve, adapt and improve. No, we are not at the age of Terminator and may never be (funny how often Skynet comes up in my discussions- you’d think it is a real network)! From the movie 2001 over forty years ago, to more recent Terminator and many other ‘machine-winning-over-human-dramafest’ movies by Hollywood, we have started to think the worse about AI. On the positive side, in the background we have robots that are evolving beautifully and the entire field of automation and robotics is growing beyond bounds, but machines are not taking over humans. Yet!

The journey of AI has begun. Right now it is filled with hype, jargon, marketing buzz and at this stage it is mostly fiction. But it is the first time that everything is geared up to go in the right direction. It is up to us to do this right. We will explore the various parts in my future articles. Some companies are doing this far better than others. Deep Minds and Google are a prime example of taking AI in the right direction. Startups are emerging that will be game changing. We are now truly at the Age of AI. I would love to hear your thoughts.

More to come on this as we get the conversation started. Disrupt | Innovate | Lead