How EI-oriented technologies will enhance human communication
by Zhenya Mirkin and Segah Mir, NLP & Big Data Experts, Caura & Co.(caura.co | @CauraNdCo)
If you are a non-native English speaker, you remember the days when you had to use an online dictionary. You would look up a word, then another — until you translated the whole sentence. This was also the case with translating emails and web pages from other languages into English.
Then, in 2006, Google launched its Translate Service. Suddenly, it was possible to translate whole sentences and even pages. The result was not perfect, but it made online dictionary services obsolete almost overnight. A few dictionary services remained popular for complex languages (Chinese Mandarin, Russian, Hebrew…), and even those are quickly losing relevance.
Twelve years later, there is another revolution brewing in the language space: Interpreting Emotions. And it is super important! To understand why, consider just how interconnected we are in today’s digital world and how much our culture of origin impacts the way we communicate. Yes, you can translate email messages word for word, or phrase for phrase, but can you truly understand the real opinions and undertones of people behind them? How do you read emotions of someone located far away? How do you tell what someone is feeling if her written or spoken English is not perfect?
What is Emotional AI?
Emotional AI is a broad range of technologies aimed at automating objective measurement of opinions, feelings, and behaviors. It relies on natural language processing (NLP) & understanding (NLU), as well as modern psychology to extract the relevant information on human opinions and feelings. Increasingly, it involves face- and voice-recognition technologies to analyze tone facial expressions and mood of the speaker.
The list of applications is vast. Consider the simplest text analysis that extracts emotional signals such as:
- Negative
- Aggressive
- Exciting
This kind of application could help writers spot problems such as excessively emotional or improperly casual vocabulary in their writing by analyzing writing style and alerting the writer to an off-sounding fragment. A more advanced application could make recommendations on how to make a sentence or paragraph sound, say, more positive. A more advanced application yet could make a recommendation on how to rewrite the text altogether to achieve a specific goal.
Of course, emotional understanding is not achieved by making every message sound positive. Context is crucial: history, relevant data, live events, and other factors contribute to how communication flows. Future applications will find and provide value through understanding how detectable signals fit in the broader context of a transaction or a relationship.
What technologies power Emotional AI?
Before talking about the types of applications one could see in the future, we would like to make an important distinction between two basic categories that shape the technology landscape:
- API services, i.e. commodity services offering well-defined basic classification (e.g. sentiment analysis, entity recognition)
- Context-aware services, i.e. value-add applications that rely on different types of data to deliver compound value.
What makes up an Emotional Intelligence AI Product?
Leading NLP API providers such as Google and IBM provide the basic general-purpose building blocks which developers compare on accuracy, price, and scalability. While this tooling may be helpful as a baseline, it is unlikely to produce significant value in any specific use case. Expecting their current offerings to be universally applicable is akin to expecting average temperature across a hospital to be useful in diagnosing the state of health for a particular patient. For that, these giants need end-user applications.
Context-aware services are optimized for well-defined use cases. This allows them to enrich applications with relevant domain knowledge and historical context: who, what, when, why. This is really where the bulk of new innovation is starting to happen. It is still a nascent space though.
Client/context-aware applications can be further distilled into categories according to the use case. Writing Productivity is the most common**.** Grammarly and Google Translate fall into this category. There are other use cases though, most notably:
Emotional AI Opportunities and Examples
On the sales enablement, automation was the SaaS playbook for the past decade. Increased capabilities in data gathering and reporting enabled sales staff to make faster and more accurate decisions as long as all data was made contextual and easy to access. The new EI opportunity essentially derives from this use case. New applications attempt to interpret communication profiles and emotional needs of prospects before any meetings even take place. Applications present this data contextually, thereby helping sales development representatives and account executives make more accurate decisions.
Like sales, customer support has undergone its own era of automation. Support agents today rely on an increasing set of integrations that help them select the right templates, reference relevant knowledge base articles, and track client’s usage. The new opportunity is essentially an extension of this into the next era, promising to automate the selection of appropriate help and resources based on emotional needs of a customer during the conversation.
Team collaboration is yet another area where EI applications are relevant. While the previous decade was about getting people connected and engaged in the right conversations, the new set of tools aims to make those engagements effective by helping with empathy and efficiency. This is not a new challenge — companies have been coaching employees on in-person communication and emails for several decades. What’s new is that instant messaging work environment (i.e. Slack) imposes a different pace and response time requirement.
With people analytics and surveys, EI applications are immersed deep into the enterprise environment, so it is hard to single-handedly describe the overall direction. One thing is clear, if the last decade focused on gathering data and reporting on that data, the new applications will aim to narrow the gap between what people self-report and how they actually behave.
Concluding Remarks
AI space is huge. It can be hard to separate futuristic ideas from applications that can be relevant today. However, we deem Emotional AI a domain ripe for real applications because the use cases are well developed and base technology is mature. The problems are quite large and will no doubt attract competition of incumbent technology companies such as Slack, Zendesk, and Hubspot. However, the technology has also finally caught up with the needs of application developers. If you pay attention to the target market (pace, needs) and available resources (tech baseline and your core competency), you should be able to make calculated bets on lucrative value-adds.
Zhenya Mirkin is a Partner and NLP Expert at Caura & Co. as well as Inventor of Entropy AI. Formerly, head of engineering at Parsable. In the early 2000s, he led the development of predictive text input software that shipped on all Motorola phones worldwide.
Segah Mir is a Principal and Big Data Expert at Caura & Co. He founded Caura & Co. in 2016, and made it one of the earliest partners for Looker Data Sciences and Google BigQuery technologies.