Arabic can be a challenging language. “It does not make sense to take a generic English sentiment, and use it for your clients who speak in Arabic,” says Dr. Zuhair Khayyat, Chief Technology Officer, Lucidya.
Lucidya is the leading platform for customer experience management in the Arab World. For the past seven years, Dr. Khayyat’s team has been working daily to collect one billion data points, and create the highest accuracy for the Arabic sentiment in the market.
Lucidya’s technology evolves with the latest research and the market. Dr Khayyat shares his perspective on the right way to use AI, the gap in the market for NLP tools that analyse the Arabic language, and how the business future lies in technologies working in sync towards a common business goal.
Excerpts from the interview
What are your favourite metrics to measure customer satisfaction, and how are these measures changing in the current landscape?
The most common metric to measure customer experience in any company is the NPS. However, this is not my preferred tool because it would require you to rely on customer responses, which may not be sufficient. Instead, use AI to predict this.
You need to look into your customer interactions across various sources and then use AI tools to predict information before something terrible happens to your reputation, or even to just know that you are going in the right direction.
Tell us about the gap in the market for NLP tools that analyse the Arabic language.
Arabic is, of course, a challenging language. The number of resources you can find online to train, analyse and create great predictions with high accuracy is very scarce. In Lucidya, we do not only focus on NLP, but we prefer NLU (Natural Language Understanding). For the past seven years, we have been working daily to collect one billion data points, and create the highest accuracy for the Arabic sentiment in the market.
Our AI models not only capture sentiment but also predict the location of the person, gender and dialect. We have high accuracy in extracting the topics that people are talking about. Even if the topics are not known previously, we try to predict what they are.
We reached 92% and are improving with the help of ML Ops and other unit testings to ensure that our models fit into the Arabic language, and accommodate different dialects and changing environments.
What advice would you give business leaders who are initiating their AI journey for CX as a function?
AI is a double edge sword. If you use it correctly, it’ll lead you to great success. But if you misuse or misunderstand it, it will lead you to failure. It would be best to use proper tools or models that fit your segment and your use case. For instance, it does not make sense to take a generic English sentiment, and use it for your clients who speak in Arabic frequently. You will get wrong predictions and wrong annotations.
Meanwhile, if it gives you unpredictable results, it means that you are missing something or there is something that you have yet to observe in your data. Keeping an open mind towards the predictions the model gives you is crucial. The team needs to understand the maths behind what’s happening in your AI models. After all, AI is just a tool, and it all comes down to how you use it.
What is your prediction for emerging technologies that will play a significant role in the CX stack?
We asked Chat GPT this question. Although it told us that it could not predict the future, it gave us helpful key pointers. For instance, it told us that AI is the leading tool to help in the customer journey, and analyse the user within his journey.
Additionally, combining other technologies like blockchain, AR, voice technology, and IoT can help create the best customer experience analysis for any business trying to perfect its image in the market.
Social Listening, Channel Analytics, Surveys, and Audience (CDP), what’s remarkable about this combination of products and what overall value does the Lucidya platform offer brands in the region?
The combination of those tools will allow any company or any client to monitor not only public channels but also other ways of communicating with customers. It’ll collect information from various social media platforms or even conversations had over email, WhatsApp, or even with your chatbot. Analysing the massive data from these sources in real-time with a single tool will allow you to understand your audience, regardless of the source.