We humans think of ourselves the most intelligent species on Earth. Our intelligence has actually made us realize that our intelligence and abilities are limited, and we need tools to accomplish certain tasks.
To begin with we copied what we saw in nature: we invented the wheel by observing a rolling boulder; we built planes by copying birds flying; we built factories and automatic machines to produce things faster.
And then we built computers to help us answer questions faster. From simple computations to the internet, the digital era opened masses of opportunities we couldn’t even think of 15 years ago.
The reality of AI
Remember how you used to search for information in your school textbooks? Page by page, paragraph by paragraph… it could take hours to find that equation. Now we have search engines that can answer almost any question like that in milliseconds.
Should we think of a search engine as ‘artificial intelligence’? Is it more intelligent than us? Well, it can definitely store more data than our brain can and ‘remember’ things much faster. But can it use the data to prove any scientific theory? Search engines just save us time, so we humans can focus on making use of that information, sticking pieces together to invent new things.
Some of us get scared when we hear about machine learning, artificial intelligence or robots, but we forget that, not long ago, we considered a calculator a very smart machine, and a toy that said ‘hello’ when you pushed a button was also a ‘smart robot’.
Nowadays, if your mobile phone hangs for 5 seconds while searching for your favorite photo in a 5000-image gallery or while answering what’s your question about what’s the best thing to do near you today, we call that phone ‘stupid’ or ‘slow’.
The reality is we are now surrounded with products and services powered by AI without even noticing it. Your phone reminds you about your upcoming flight by just reading your email confirmation from your airline company; you are notified when something you are interested into has just arrived in stock; and you get alerts for bad weather.
The capabilities and limitations of AI
We tried to copy the way our brain works, so we built artificial neural networks that can learn things and make decisions. That worked… for very basic tasks.
Not until recently, when we humans managed to build much faster computers, did we manage to solve more complex tasks with deep learning. Today we can now properly extract information from huge data sets in real time, we can recognize patterns in disease occurrences, we can process signals from millions of IoT devices, and we can even answer questions about the universe.
But it’s us, the humans, we’re the ones that define those complex tasks, the ones that use deep-learning solutions to draw conclusions, and it’s us that can then move further in defining even more complex tasks.
Let’s consider an example. Would you let a robot perform a surgery based on a decision a computer has made from your blood tests and your MRI? Or would you be more confident if a doctor also validates those decisions and maybe asks for extra information the computer hasn’t been taught to consider? The robot and the computer are there to help avoid human errors and to compare your case with millions of other cases a doctor can never personally see in his lifetime.
We haven’t reached that level where AI bypasses human intelligence. It’s the combined, collaborative intelligence of these two that takes our lives to the next level.
We now have self-driving cars that can analyze traffic and the environment much faster than a human can. The car can successfully avoid crashes by using video cameras, lasers and AI to process all those signals. But what would the car do if your beloved dog appears suddenly in front of your car running after a cat? Will the car decide it’s safer to crash into your garbage can rather than hitting your dog?
We still need humans to put things into the right perspective, to add more context and to apply their intuition.
Chat bots and voice assistants
The digital revolution brought new types of human-computer interactions. We’ve learned to read from a screen instead of a book; we are using a mouse or touch to navigate; and we’re using a keyboard to type. We have adapted so much that we now prefer to send chat messages to our friends instead of calling them. That’s why we are now building smart chat bots that can help us with almost any matter.
Even if chat bots can help you answer frequent questions, schedule meetings, solve complex IT tickets or automate almost every repetitive task you deal with day by day, they’re not pretending to be humans. There have been multiple attempts to pass the famous Turing Test, but all of them have shown how hard to achieve that is. Chat bots are not trying to fool people they are real people, and they usually start each conversation with a list of limitations or skills.
We are now beginning a new chapter where chat bots are being replaced with voice assistants like Alexa or Siri. After years of typing and moving our lives to text-based social networks, we are rediscovering how natural it feels to just speak when you want something. We are amazed how easy it feels for a toddler to interact with Alexa.
There have been social experiments where voice assistants have successfully improved the lives of the elderly, or have helped cure depression or anxiety even though people were aware they were not talking with a real person.
Artificial intelligence is finally blending nicely into our lives. While it can automate certain tasks and might put some low-paying jobs in danger, it also heralds new horizons, generates new industries, creates new and more interesting jobs, and is helping us to evolve into an even more intelligent species.