“The greatest danger in times of turbulence is not the turbulence; it is to act with yesterday’s logic.” - Peter Drucker
…
At this point, you’ve likely been inundated with with the news of AI, LLM, GPT, or ChatGPT. Hint: GPT stands for "Generative Pre-trained Transformer". And it likely begs a few questions… Is this the start of SkyNet? Will I become irrelevant? And a multitude of others. This anxiety is OK and is a good thing; it alerts us that a fundamental shift is happening, just like the internet had done previously. And, no, this isn’t the beginning of SkyNet.
Let’s face it; support is moving to a more asynchronous communication model (a-la chat). With the ability for ChatGPT to “stream” messages similar to how a user would, backed by the knowledge of the internet, the result can be a very powerful one-two punch. Plus, it can eliminate a lot of the “grunt work” (aka work “suck”) that support engineers commonly deal with, allowing them to focus on the real problems at hand.
Also, with this evolution to a semi-sentient system, the experience for end-users is far superior, sometimes to the point where you may not even know you’re conversing with a computer.
Benefits
Here are some of the potential benefits:
Efficient Response Times
- AI-powered chatbots can handle multiple customer queries simultaneously, significantly reducing response times and providing instant support, resulting in higher customer satisfaction levels.
- By offloading the initial responses to an intelligent system, you can buffer the amount of time a traditional agent has to respond
- This may also help ensure you meet you SLA requirements
Cost-Effectiveness
- AI-driven customer support tools can minimize the need for large customer service teams, reducing operational costs without compromising on the quality of support (this needs to be validated)
- You can make users more productive by offloading some context gathering and/or successfully deflecting certain queries
- Any gain in productivity may minimize the need to scale team sizes up, and may actually allow you to handle more with a smaller team
24/7 Availability
- AI-enabled customer support systems can operate round-the-clock (they don’t need sleep), ensuring customers receive assistance whenever they need it, irrespective of geographical and time zone differences.
- For a follow-the-sun model this may be less important, however, if you don’t have this coverage in place, an AI-assist can be extremely valuable to provide coverage between gaps
Personalization
- By leveraging natural language processing and machine learning, ChatGPT and similar tools can understand customers’ needs and preferences, providing personalized and contextually relevant support.
- If the vendor provides the tool with additional context (e.g., about the user, their use-cases, etc.), the system can provide even more personalized information
Continuous Improvement (‘fo free)
- AI-driven support systems can learn from customer interactions, adapting and improving over time, resulting in better support experiences and higher customer retention.
- Traditionally things would be done in a rule-like manner, which was inflexible and required intervention on every change.
- Just like any service, as compared to build your own, the service can evolve and improve, allowing you, and your customers, to reap the benefits
Multilingual Support
- AI solutions can understand and respond in multiple languages, breaking down language barriers and allowing businesses to cater to a diverse customer base.
- This may negate the need for traditional translation services, as the service can handle this natively
Potential Risks
While the use of artificial intelligence and tools like ChatGPT offers numerous benefits in the realm of customer support, it is crucial to consider potential risks that may arise.
Some of these risks include:
Misunderstandings
- AI-driven chatbots might misunderstand complex customer queries or interpret context inaccurately, leading to incorrect or irrelevant responses. This can result in customer frustration and diminished satisfaction.
- While this is likely a edge case, you can easily account for this with logic in the system than can allow the user or system to automatically switch the interaction to a support engineer.
Lack of empathy
- Chatbots may struggle to replicate the empathy and human touch that customer service representatives provide. In sensitive or emotionally charged situations, an AI-driven response may appear impersonal or unsympathetic, negatively impacting customer relations.
- This is both a good and bad thing as lack of emotion allows them to stay calm when customers may be spicy, however, empathy is one of the most important things to show customers as we “feel their pain”.
Over-reliance on automation
- Businesses might become overly reliant on AI-driven solutions, compromising the quality of customer support. While AI is highly efficient, it cannot fully replace the nuanced understanding and problem-solving skills of human agents, especially for complex or unique situations.
- A great example of this is knowledge with the introduction of the internet. Because we are not forced to retain knowledge, we can just store it ephemerally and query when we need information.
Data privacy concerns
- AI systems rely on vast amounts of data to operate effectively. The collection, storage, and processing of customer data can raise privacy concerns and expose businesses to potential data breaches or misuse of information.
- This will be a very significant area of focus and will likely impact where AI-assists can be used.
Bias and discrimination
- AI systems may inadvertently perpetuate biases present in their training data. This can lead to biased or discriminatory responses, negatively affecting customer experiences and causing reputational harm to the business.
Ethical concerns
- The use of AI in customer support raises ethical questions about job displacement, accountability, and transparency. Businesses must carefully weigh the benefits against potential social and ethical consequences.
- However, at the end of the day, the businesses goal is to operate as optimially as possible, and provide the best possible support.
- If this is possible with the displacement of positions, while ethically questionable, it may make sense from a fiduciary perspective.
How To Embrace
Luckily a lot of software vendors have hopped on the “GPT train” announcing AI-assists in their products. The following is a list is some that are relevant in the support world:
- Intercom (Wingman)
- DevRev (Turing)
- SFDC (Einstein)
The other option is to leverage the OpenAI API and create your own integration. With building your own integration you get the benefit of flexibility and choice with the tradeoff of time, cost and maintenance.
One key for any of these is to ensure you validate and potentially modify the responses before sending to your customer. You can either do this on query (e.g. feed in your taxonomy), by post-processing the result, or both (preferred).
Outro
The application of artificial intelligence and tools like ChatGPT in customer support can drive significant improvements in response times, cost-efficiency, personalization, and overall customer satisfaction. Embracing these technologies will not only help businesses stay competitive but also foster long-term customer loyalty (and employee retention).
However, to mitigate these risks, businesses should adopt a balanced approach that combines AI-driven solutions with human support. Continuous monitoring, improvements, and ethical considerations must be prioritized to ensure AI’s responsible and effective use in customer support.
TL;DR
- AI takes the support experience to a whole new level
- Embrace it
- Key is finding the balance between autonomy and human control
- Like all things, start small, and increment