When You Will Encounter Genuine Emotional AI

Genuine Emotional AI

Recent tremendous growth of mobile technologies and advancements of artificial intelligence (AI) in virtually any industry will give a boost to some quite unexpected applications of AI in the coming decades. Emotional AI is one of the fields where machine learning and algorithms offer huge potential although reading and controlling emotions through AI is quite a challenging task. At this stage, researchers focus on developing AI algorithms that can read and respond to human emotions. Once we have viable technology that accurately interprets our emotional status and reaction, we can start thinking of building AI-powered systems that are capable of convincingly simulating emotions or have their own emotions.

Reading Emotions and Reacting to Them

Possible applications of emotional AI spans areas ranging from self-driving cars, to personal robots and virtual assistants, to IoT computing and literary all major consumer electronic devices. Any of those devices or apps could utilize AI to provide enhanced human-machine interaction, emotional awareness or machine emotional intimacy. This is also a way to overcome prejudices related to use of AI-powered systems as a reliable tools assisting humans in their personal and work life. In fact, development of emotionally intelligent devices and software is a crucial factor for the overall growth of AI applications across any industry.

Androids should not the ultimate goal; one would reasonably argue that having human-like robots does not make sense by any means. Nevertheless, a machine-learning algorithm coupled with AI capabilities can be extremely useful in human-machine interactions. American psychologist Paul Ekman identified six basic emotions in the 1980s − anger, disgust, fear, happiness, sadness, and surprise. Later, Professor Robert Plutchik extended basic emotions to eight, grouping them into four pairs of polar opposites: joy-sadness, anger-fear, trust-distrust, and surprise-anticipation.

A few AI startups work on development of algorithms that are able to recognize these human emotions and eventually develop similar emotions in intelligent machines. Although founders of emotional AI startups like Patrick Levy-Rosenthal claim their algorithms are achieving up to 98 percent emotion recognition accuracy on conversations we still do not have viable emotional AI for widespread use. Facial recognition of emotional states is progressing rapidly but most human-machine interactions are performed without the use of a camera. Hence, natural language understanding appears as a more crucial technology to develop if we want to build viable solutions utilizing emotional AI.

How to Trust Emotional AI?

Making a device that develops unique personality through reinforcement learning and Machine Learning is achievable today. Nonetheless, it has applications mostly as an entertaining device unless it is able to recognize human emotions with 100 percent accuracy and take meaningful actions depending not solely only on emotions but while taking into account a complex set of other factors.

Imagine an autonomous vehicle that can also be controlled using voice commands. The vehicle is speeding to a crosswalk where a pedestrian is crossing the road. An emotional and instinctive reaction of the driver would be to shout “Stop” or give a similar command. What if the better decision is to speed up and find a way around the left or the right side of the pedestrian? How should the car’s AI react to the emotional voice command as opposed to its algorithms? Obviously, a priority of computed algorithmic actions can be programmed for such a situation but the above case is a relatively simple one. What if the driver/passenger experiences a momentary lapse of reason and issues a command that has nothing to do with reality and immediate surrounding? Emotion recognition and interpretation is much more complicated than simply reading facial and language expressions.

Nonetheless, emotional AI is progressing rapidly and Alexei Samsonovich, a professor in the Cybernetics Department at the Moscow Engineering Physics Institute, already proposed a multi-part test intended to answer a critical question: How do we know that such intelligence actually experiences real, human-like emotions?

Virtual agents and robots should be human-like so that humans could trust them and cooperate with them as with their equals. Therefore, artificial intelligence must be socially and emotionally responsive and able to think and learn like humans. And that implies such mechanisms as narrative thinking, autonomous goal setting, creative reinterpreting, active learning, and the ability to generate emotions and maintain interpersonal relationships,” Samsonovich explains.

Prepare for Emotional AI

Creation of such a test is possible in a completely virtual environment where no face-to-face contact is available. However, emotional AI also faces a multi-attribute classification problem related to facial expressions. Human facial expressions, for instance, involve eyes, brow, and lips but also depend on age, race, and gender. AI researchers and developers should also take into account factors like lighting and orientation to the camera.

Body movements i.e. body language is even more challenging for AI to interpret. Humans have learned to decipher body language through millions of years of evolution. However, it is not simply about reading body language but putting it into context. Shrugging your shoulders is a common gesture but it could express different emotions in different contexts.

Finally, you’ll have to wait for an android to join your family for the time being, but a mass device that is capable to understand basic human emotions and use AI to respond and take actions based on them will be available somewhere around 2020.

By Kiril V. Kirilov

Rick Braddy

The Secrets to Achieving Cloud File Storage Performance Goals

Storage Performance with Cost Reduction By 2025, according to Gartner, 80 percent of enterprises will shut down their traditional data centers. As of 2019, 10 percent have already shifted their data centers and storage to ...
Alex Brisbourne

Industrial IoT Cyberattacks Continue To Rise

IoT Industrial Security The Internet of Things (IoT) includes both traditional electronics and everyday ‘things’ embedded with sensors, computing, and networking capabilities. From smart coffee makers and smart homes to smart lighting and smart cities, ...
Nik Thumma Contributor

Why It’s Time for Companies to Move ‘All-In’ on the Cloud

Companies to Move ‘All-In’ on the Cloud The cloud offers businesses innovative ways to optimize operations and achieve amazing results. While many companies have already migrated to the cloud in some capacity, the full scope ...
Will Crump

The Key to a Successful M&A = Data

Successful M&A = Data Data is often the single point of failure for many organizations. Divestitures, privatization, leveraged buyouts, and management buyouts are all on the rise, but data too often remains an afterthought, rather ...
Mobile Apps Business

It May Not Be Sexy, But Strict Compliance Delivers The Freedom To Innovate

Compliance and Business Innovation When the U.S. based non-profit organization RHD | Resources for Human Development decided to move its operations into the cloud, one of its top priorities was compliance. As a company that ...
Wasabi

Episode 3: The Bottomless Cloud – An Interview with David Friend of Wasabi

Why data is not “the new oil” and why “cloud” means more than we think. In his new book, author David Friend refers to the cloud as "bottomless," and disputes peoples' assessment that data is ...