By Nigel Cannings, Founder, Intelligent Voice
The integration of voice AI into marketing and customer service holds the potential to provide businesses with more comprehensive customer profiling, improved data collection, and better data analysis. However, there is an apprehension felt amongst both customers and businesses concerning the expanded implementation of recording software, despite the potential benefits to both parties. It is important to establish an understanding of the legalities surrounding consent to record, how customer profiling works, and the flaws in current AI profiling systems.
What does the rise in voice recording mean for businesses and customers?
Increased application of voice recording will result in customers experiencing a higher volume of interactions prefaced with automated messaging systems announcing the recording of calls for various purposes. Although sometimes alarming, voice recording is significantly contributing to the improvement of customer experiences.
Voice recording brings several benefits to businesses. AI-orientated recording presents businesses with the means to compile detailed customer profiles which can further improve and personalise customer interactions. Customers can receive targeted information and sales drives, avoiding unnecessary marketing material that may result in disengagement or a negative brand image. Voice recording is also rapidly becoming an important resource for compliance monitoring, demonstrating a company’s cooperation with regulatory measures.
How does voice recording and AI profiling work?
AI profiling and voice recording is usually founded on Conversational AI, Natural Language Processing (NLP), and Automatic Speech Recognition (ASR). These functions collaborate to provide machine learning systems with adequate data to configure algorithms using their recorded history, improving the functioning of the algorithm and overall system. This ensures that the data collected from recorded calls can reach its full potential, better informing marketing decisions.
Customer profiling AI collects data on speech characteristics, language features, and behavioural indicators, forming a more comprehensive understanding of each customer. This information builds detailed customer profiles that can inform business decisions and customer interactions. They can include information about a customer’s conversational style, attitude towards business interactions, and mood. As a result, customers identified as being of a certain nature or requiring a more specific customer service experience can be connected to the call handler with the most appropriate skill set. Through this, companies can increase their sales potential and create an overall more positive customer experience.
AI is also providing businesses with a more accurate and comprehensive overview of their customer base, examining their customers’ attitude towards their brand image, the reactions they are receiving from customer service interactions, and the customer response to promotional information and advertising. It has been estimated by Forbes that voice-based AI systems will be handling 20% of all customer service queries by 2022, bringing more businesses the benefits of utilising these systems.
What is the risk of existing bias in AI?
As a new and still developing technology, there have been issues identified in the implementation of AI profiling, with the most notable being the presence of cultural biases. There have been attempts to use AI customer profiling to identify physical features through the analysis of voice data, including an individual’s height, weight, gender, or health status – these attempts have frequently led to the detection of race or gender bias. Speech recognition functions have also been identified to hold these biases, performing more inaccurately for women and non-white individuals. These biases are not intentionally coded into speech recognition software, but are nevertheless problematic, and will likely require more specific research and coding to correct.
It is important for businesses wishing to implement this software to note that these difficulties faced by AI profiling can lead to the collection of skewed data, therefore risking the recording of incorrect customer information and inaccurate analysis of the resulting data.
How should businesses manage consent for AI-based voice recording?
There are already various legal regulations in place for businesses using voice recording technology, especially in customer-facing interactions. Two primary forms of consent may be required in this context: one-party and two-party consent. One-party consent refers to where only one person being recorded needs to be aware of the recording taking place. Two-party consent means that all parties must be informed and consenting to be recorded. It is commonplace to use automated messaging systems to inform the presence of recording prior to commencing customer calls.
This ensures that both parties are aware and consent to the use of data. Although the increased presence of voice recording may feel initially invasive, these recordings provide customer protection and hold businesses to the correct regulations for safety and security.
