Qual é o problema do ChatGPT no contact center?

Explorando as preocupações éticas do ChatGPT no contact center.

I’ve been writing feverishly lately about the opportunity ChatGPT (and other generative AI systems) presents for the contact center. At Talkdesk, we believe that the generative pre-trained transformer (GPT) large language models (LLMs) powering systems like ChatGPT are going to be the backbone of the next generation of contact centers as a service (CCaaS).

We’ve launched our first feature that harnesses these capabilities to automatically summarize customer conversations and accurately select their dispositions (e.g., requests follow-up, wants to cancel service, etc.)—effectively eliminating most of a customer service agent’s after call work. And we are actively integrating these LLMs into other core contact center use cases, as well as shaping our roadmap to accelerate the value our customers can gain from these natural language processing techniques.

Mas também compreendemos a magnitude desta nova tecnologia e o potencial que tem para causar tanto danos como benefícios.

A number of prominent tech leaders—including Elon Musk, Steve Wozniak, and Andrew Yang—are publicly calling for a pause in the training of AI systems more powerful than ChatGPT-4. Their general ethical concern is that runaway AI will overtake humanity’s ability to harness it for good.

Com esses grandes nomes avaliando alguns pontos negativos dos sistemas de IA, pensei que seria um bom momento para delinear algumas das principais preocupações éticas desses novos LLMs para o contact center. Qualquer empresa que considere usar esses modelos de IA em seu contact center deve estar ciente das possíveis armadilhas e considerar processos que ajudarão a garantir experiências seguras e positivas para os clientes.

Cinco preocupações éticas do ChatGPT no contact center.

1. Transparência.

The first ethical concern with using ChatGPT and other generative AI systems in the contact center is transparency, which applies both to being clear about how an AI-driven decision was made as well as the fact that it was made by AI in the first place. Customers have the right to know if their interaction was mediated by AI. If customers believe they are talking to a human and later discover that they were talking to a chatbot, for example, it could erode trust in the company and damage customer relationships with the brand. Full disclosure to the customer is crucial when designing chatbots and virtual agents.

2. Preconceito.

Another ethical concern is the potential for bias in the responses of chatbots and virtual agents. ChatGPT is a machine learning model that is trained on vast amounts of text data, and there is a risk that this data could be biased in some way. If, for example, an automatic quality review favors one pattern of speech over another, this could lead to unequal treatment of agents whose backgrounds included learning that pattern of speech. In other words, this type of bias can lead to discrimination.

3. Segurança e privacidade de dados.

When using ChatGPT, user interactions are logged and stored, and personal conversations could be accessible to anyone if proper security measures are not in place. As with any software system handling personally identifiable data, data flowing through the LLM needs to remain encrypted end to end and follow all of the data privacy considerations that keep customers’ data safe.

4. Resultado verdadeiro.

Outra preocupação ética é o potencial do ChatGPT “alucinar” ou inventar respostas que podem não ser verdadeiras, mas transmitir esses fatos alternativos de uma forma convincente. É claro que não queremos dar informações erradas aos clientes, especialmente se eles as usarem para tomar decisões na vida.

5. Deslocamento de emprego.

Finally, there is the concern of job displacement. As more companies turn to AI-based chatbots like ChatGPT, there is a risk that human contact center agents could lose their jobs. Companies must consider the impact that automation could have on their employees and take steps to mitigate any negative effects. In particular, contact center managers should imagine new opportunities to increase agent satisfaction by introducing tools that will automate menial tasks and help agents complete their work faster.