Prelude: Is AI Getting More Human?
50 years ago, legendary director Stanley Kubrick introduced the world to HAL 9000, the autonomous killer A.I. from 2001: A Space Odyssey. Then, we laughed at the idea that we would ever live in a world where a computer could match or even exceed the intelligence of humans.
Today, we’re not laughing anymore. We now live in a world wherein A.I. is as ubiquitous in our daily lives as our reliance on the devices they power. No more are we worried about the intelligence of our computers– today, we’re worried that we won’t be able to tell the difference between A.I. and humans.
“The Uncanny Valley”
That’s the name given by Masahiro Mori in 1970 to the experience we feel when we see something that acts almost like a human but something about it just isn’t quite right. While this feeling is most common in the physical world – for say, when you see a robot that has a human face but inhuman expressions – this feeling can also come from computer animations or even conversations over the phone with an A.I bot.
Yet, our inquisitive nature demands progress and so, current and future research in A.I. means that computers will become ever smarter and A.I. ever more human-like. With this inevitability, if we humans cannot beat back that progress, we can work together to make A.I. human enough that we no longer experience the Uncanny Valley effect.
With that in mind, let’s take a look at the top trends to watch in humanizing A.I. in 2019.
Trend #1: The Adoption of AI In Our Daily Lives
2019 is the year of artificial intelligence. It may be only January but A.I. is starting the year with a splash. Once a buzzword used only exclusively by techies and technology first adopters, after the year that A.I. has had, 2019 is the year that A.I. really becomes a part of our every day lives.
Every year, people turn to the Las Vegas Consumer Electronics Show to see what are the trends and new products that companies are investing in. In 2019, A.I. is going to be a major theme at the CES, showcasing in a variety of products from home products to entertainment to even sports equipment.
And it is not just CES where A.I. is going to play a major role. From fighting hackers to protecting humanity from itself, this year is shaping out to be the year where A.I. moves from the laboratories of universities to becoming as common a system in your home as your bed, think I, Robot common.
Trend #2: AI x Blockchain
The Internet of Things, Artificial Intelligence and blockchain are but some of the buzzwords that have been thrown around. If the past few years were being used to develop the technology behind those buzzwords, 2019 is the year when those technologies will converge and work with each other instead of operating in independent silos.
In as technologically reliant a world as we live in now, we generate massive amounts of data. On average, Internet users generate 2.5 quintillion bytes of data every day (that’s 2.5 billion billion bytes). In fact, we generate so much data that 90% of the world’s data was generated in the past two years alone!
To parse through all this data, scientists use a method called Deep Learning to train their Artificial Intelligence systems. The problem however, is that most of the time, we don’t know what our A.I. is doing most of the time. Here’s where we note the trend of converging the technologies between Artificial Intelligence and blockchain–there are already startups doing that– such as Connectome, which builds virtual human agents. If blockchain is used to keep an indelible record of processes made and we struggle to understand what A.I. do because we don’t have a good record of what they do, then using a GeneFlow system allows one to create that record.
GeneFlow refers to a method whereby the traceability of an A.I.’s learning and execution history is realized in a blockchain. When an A.I.’s processes an be reliably traced, it allows for a chain of accountability to emerge, making the A.I. more reliable because we can see into the blackbox that is its decision making process.
Trend #3: A.I. To Sound More Human
If there is one thing that makes HAL 9000 so incredibly scary, it’s that HAL talks to us like how we talk to each other.
In this famous movie scene, HAL says, “I’m sorry Dave, I’m afraid I can’t do that,” because HAL knows that Dave is trying to disconnect it. In 1968, to listen to a computer respond and rationalize their actions was only possible in the realm of science fiction. In 2019, it’s an every day occurrence. Siri and Alexa aren’t just names of people anymore. They’re the names we use when we want to activate our Apple, Amazon or Android assistants.
However, unlike the movie, our A.I. agents do not just randomly slap words together to make a sentence. They require massive data sets to be used to train them. That is why tech companies around the world are now hiring not just computer engineers but social scientists and linguists to help them make their A.I.’s more human. To make them more human, it is both about how they sound and how they feel. When you can make your A.I more human, not only do they help in banishing the Uncanny Valley effect, you make your A.I more trustworthy and relatable.
It is that feeling of relatability and trustworthiness that makes us want to use these A.I more. When we can relate to them, we become more comfortable speaking aloud to them. When we can rely on them like a secret invisible friend, we know that our secrets are safe (mostly).
To do that though, we need to have them start sounding more human. We’re all familiar with how robotic a, well, robot sounds. One of the most famous voices in the world, the late Stephen Hawking, used a voice synthesizer and that is one of the first images that come to mind when we think about a robotic voice.
Technology has improved to the point that our virtual human agents now sound human only because their voices were made from voice clips recorded by actual humans. However, because there are only so many clips that are available, additional phrases are often synthesized by cutting the relevant words and putting those clips together. Sometimes that method words, often, it’s a mess. However, we are near that goal of realistic sounding voice, as Google just recently showcased, and 2019 is that year when we cross the finish line.
A.I., Humanized: Where Do We Go From Here?
From companies like Google and Amazon to the US Army, Artificial Intelligence is here to stay. As our smartphones get smarter, computers faster and virtual agents more common, 2019 will showcase the power of artificial intelligence like never before.
However, as the Uncanny Valley effect shows, there is a danger that if A.I. becomes more robotic and less human, regardless of how powerful a tool it becomes, we will still reject it. Technological convergence in making A.I. more accountable and creating A.I. that sounds and feels like a human will be a challenge.
And that makes 2019 all the more exciting.
e27 publishes relevant guest contributions from the community. Share your honest opinions and expert knowledge by submitting your content here.