Tech giants are racing to establish their presence, and VCs are pouring money wildly, all in a frenzy to enable "AI mind reading."

Author: Lexie

Editor: Lu

In the grand discussion about AI, people assign it roles that are either our most efficient assistants or the "machine legion" that will disrupt us. Whether friend or foe, AI not only needs to complete tasks assigned by humans but also to "understand" human hearts. This ability to read minds has been a major focus in the AI field this year.

In the emerging technology research report on enterprise SaaS released by PitchBook this year, "Emotion AI" has become a significant highlight. It refers to the use of emotional computing and artificial intelligence technologies to perceive, understand, and interact with human emotions. By analyzing text, facial expressions, voice, and other physiological signals, it attempts to understand human emotions. In simple terms, Emotion AI aims for machines to "read" emotions like humans, or even better.

Its main technologies include:

• Facial Expression Analysis: Detecting micro-expressions and facial muscle movements through cameras, computer vision, and deep learning.

• Voice Analysis: Recognizing emotional states through voiceprints, tone, and rhythm.

• Text Analysis: Interpreting sentences and context using natural language processing (NLP) technology.

• Physiological Signal Monitoring: Analyzing heart rate, skin response, etc., using wearable devices to enhance interaction personalization and emotional richness.

Emotion AI

Emotion AI evolved from sentiment analysis technology, which primarily analyzes interactions through text, such as extracting user emotions from social media. With the support of AI, integrating various input methods like visual and audio, Emotion AI promises more accurate and comprehensive emotional analysis.

01 VC Investment, Startups Securing Huge Funding

Silicon Rabbit observes that the potential of Emotion AI has attracted the attention of many investors. Some startups focused on this field, like Uniphore and MorphCast, have already secured significant investments in this track.

Based in California, Uniphore has been exploring automated conversation solutions for enterprises since 2008. It has developed multiple product lines, including U-Self Serve, U-Assist, U-Capture, and U-Analyze, helping clients achieve more personalized and emotionally rich interactions through voice, text, visual, and emotion AI technologies. U-Self Serve focuses on accurately identifying emotions and tones in conversations, enabling businesses to provide more personalized services to enhance user engagement satisfaction;

U-Self Serve

U-Assist improves customer service agents' efficiency through real-time guidance and workflow automation; U-Capture provides deep insights into customer needs and satisfaction through automated emotional data collection and analysis; and U-Analyze helps clients identify key trends and emotional changes in interactions, offering data-driven decision support to enhance brand loyalty.

Uniphore's technology is not just about making machines understand language; it aims for them to capture and interpret the emotions hidden behind tone and expression during human interactions. This capability allows businesses to interact with customers not merely mechanically but to better meet their emotional needs. By using Uniphore, companies can achieve an 87% user satisfaction rate and a 30% improvement in customer service performance.

To date, Uniphore has completed over $620 million in funding, with its latest round of investment being $400 million led by NEA in 2022, with existing investors like March Capital also participating, bringing its post-round valuation to $2.5 billion.

Uniphore

Hume AI has launched the world's first empathetic voice AI, founded by former Google scientist Alan Cowen, who is known for pioneering semantic space theory. This theory understands emotional experiences and expressions by revealing the nuances of voice, facial expressions, and gestures. Cowen's research has been published in numerous journals, including "Nature" and "Trends in Cognitive Sciences," covering the most extensive and diverse range of emotional samples to date.

Driven by this research, Hume developed a conversational voice API - EVI, which combines large language models and empathy algorithms to deeply understand and analyze human emotional states. It can not only recognize emotions in speech but also respond with more nuanced and personalized reactions during interactions, allowing developers to use these features with just a few lines of code and integrate them into any application.

Hume AI

One of the main limitations of most current AI systems is that their instructions are primarily given by humans. These instructions and prompts are prone to errors and fail to tap into the vast potential of artificial intelligence. Hume's empathetic large language model (eLLM) can adjust its choice of words and tone based on context and the user's emotional expressions. By prioritizing human happiness as the first principle for machine learning, adjustment, and interaction, it can provide users with a more natural and authentic experience in various scenarios, including mental health, education training, emergency calls, and brand analysis.

In March of this year, Hume AI completed a $50 million Series B funding round led by EQT Ventures, with investors including Union Square Ventures, Nat Friedman & Daniel Gross, Metaplanet, and Northwell Holdings.

Another player in this field is Entropik, which specializes in measuring consumer cognition and emotional responses. Through Decode, a function that integrates the powers of emotion AI, behavior AI, generative AI, and predictive AI, it can better understand consumer behavior and preferences, thus providing more personalized marketing recommendations. Entropik recently completed a $25 million Series B funding round in February 2023, with investors including SIG Venture Capital and Bessemer Venture Partners.

Entropik

02 Giants Involved, A Battleground

Tech giants are also making moves in the Emotion AI field, leveraging their advantages.

This includes Microsoft Azure's Cognitive Services Emotion API, which can identify various emotions such as joy, anger, sadness, and surprise in images and videos by analyzing facial expressions and emotions;

IBM Watson's Natural Language Understanding API can process large amounts of text data to identify underlying emotional tendencies (such as positive, negative, or neutral) for more accurate interpretation of user intent;

Google Cloud AI's Cloud Vision API has powerful image analysis capabilities, quickly recognizing emotional expressions in images and supporting text recognition and emotional associations;

AWS's Rekognition can also detect emotions, recognize facial features, and track changes in expressions, and can be combined with other AWS services to create a complete social media analysis or emotion AI-driven marketing application.

Cloud Vision API

Some startups are moving faster in the Emotion AI field, to the point where tech giants are looking to "poach" talent. For instance, the unicorn Inflection AI caught the attention of investor Microsoft for its AI team and models. After Microsoft, along with Bill Gates, Eric Schmidt, and NVIDIA, invested $1.3 billion in Inflection AI, it extended an olive branch to Mustafa Suleyman, one of the co-founders and a leader in AI. Subsequently, Suleyman and over 70 employees transitioned to Microsoft, which cost the company nearly $650 million.

However, Inflection AI quickly regrouped, forming a new team with backgrounds in Google Translate, AI consulting, and AR, continuing to focus on its core product, Pi. Pi is a personal assistant capable of understanding and responding to user emotions. Unlike traditional AI, Pi emphasizes building emotional connections with users, perceiving emotions through voice, text, and other inputs, and demonstrating empathy in conversations. Inflection AI views Pi as a coach, confidant, listener, and creative partner, rather than just a simple AI assistant. Additionally, Pi has a powerful memory feature that allows it to remember users' multiple conversation histories, enhancing the continuity and personalization of interactions.

Inflection AI Pi

03 Development Path, Coexistence of Attention and Doubt

While Emotion AI embodies our expectations for more humanized interaction methods, like all AI technologies, its promotion is accompanied by attention and skepticism. First, can Emotion AI truly interpret human emotions accurately? Theoretically, this technology can enrich the experience of services, devices, and technologies, but from a practical perspective, human emotions are inherently vague and subjective. As early as 2019, researchers questioned this technology, stating that facial expressions do not reliably reflect true human emotions. Therefore, relying solely on machines to simulate human facial expressions, body language, and tone to understand emotions has certain limitations.

Secondly, strict regulatory oversight has always been a stumbling block on the path of AI development. For example, the EU's AI Act prohibits the use of computer vision emotion detection systems in fields like education, which may limit the promotion of certain Emotion AI solutions. States like Illinois in the U.S. also have laws prohibiting the collection of biometric data without permission, directly restricting the prerequisites for using certain Emotion AI technologies. At the same time, data privacy and protection are significant issues. Emotion AI is often applied in fields like education, health, and insurance, which have particularly strict data privacy requirements. Therefore, ensuring the security and lawful use of emotional data is a challenge that every Emotion AI company must face.

Finally, communication and emotional interpretation between people from different cultural regions are challenging, and this poses a test for AI. For instance, the understanding and expression of emotions vary across different regions, which may affect the effectiveness and completeness of Emotion AI systems. Additionally, Emotion AI may face considerable difficulties in addressing biases related to race, gender, and gender identity.

Emotion AI promises not only to reduce labor efficiently but also to be considerate in reading minds. However, can it truly become a universal solution for human interaction, or will it end up being just another smart assistant like Siri, performing mediocrely in tasks that require genuine emotional understanding? Perhaps in the future, AI's "mind-reading" capabilities will revolutionize human-machine and even human interactions, but at least for now, truly understanding and responding to human emotions may still require human involvement and caution.