Insight from Dan O’Connell, chief AI officer, Dialpad
When we think of the word ‘hallucinations,’ one’s mind may conjure images of people seeing things that aren’t there. With artificial intelligence (AI), it’s very similar, hallucinations occur when machines generate incorrect or unrealistic information.
AI hallucinations are an interesting phenomenon that highlight the complexities of machine learning algorithms and the ongoing quest for AI systems to interpret information accurately. It becomes mission critical for businesses who use this type of technology, particularly in customer communications, to ensure the appropriate tone of voice is used, the communication style is on brand and the correct information is shared. This begins with making sure that the data that the AI is trained on is continually ‘parented’.
AI adoption is set to soar. Currently 15% of businesses are using AI technology and the UK chancellor is encouraging more AI adoption to boost the economy. As AI adoption grows, organisations must educate themselves on parenting this technology so that the impact remains positive. ‘AI parenting’, or the practice of humans always having oversight of their AI, can help businesses avoid potential incorrect information being shared with the public. At a time when AI is being used for many job functions, businesses cannot afford to neglect their human oversight.
Don’t keep humans out of the loop
One thing that chatbots such as ChatGPT do not allow for is transparency and control over the knowledge it draws on. This means the ability to see the facts, tone of voice and information that it will use to create content before it is delivered to end users. By screening AI responses before they reach customers, chatbots can continuously learn and improve from each interaction.
Without proper parenting, the chances of generating incorrect or inappropriate responses can increase. This can result in negative customer experiences or significant brand damage – both reputationally damning. Examples of such experiences include Google’s Bard chatbot producing the wrong answer in a promotional video, sending company shares plummeting. There were also doctors warning against using ChatGPT for medical advice after it answered questions around illnesses incorrectly. These experiences highlight the need for humans to remain in the loop to help maintain control and transparency of AI systems.
AI parenting can also play a vital role in addressing bias. By actively involving humans in the process, biases can be identified and corrected, avoiding potential controversies. Instances of AI discrimination against individuals based on race and other factors have garnered significant media attention, underlining the need to uphold human values and prevent personal bias from influencing AI decision making. Failure to do so could lead to severe reputational damage and legal implications.
The importance of AI parenting for organisations
AI parenting plays a crucial role in addressing the challenges AI systems face, such as bias, lack of contextual understanding and inability to handle complex situations effectively. This transparency creates a sense of trust and confidence, as customers know that their concerns are being acknowledged and addressed with the human touch.
To ensure a loyal customer relationship, the adoption of AI parenting practices should be considered by all companies that use AI. By incorporating human oversight into AI interactions, businesses can demonstrate their commitment to accuracy and reliability, and assure customers that there are safeguards in place to manage their sensitive data.
Consumers value digital trust and want to have confidence that companies are protecting their privacy – with 72% of respondents to a recent McKinsey survey claiming that knowing a business’ AI policy is important before making a purchase. As well as trust, parenting of your AI systems can also hugely help with growing customer engagement as the data collected through customer interactions can be used to personalise the customer experience, through making unique recommendations based on customer behaviour. AI can be used to understand customer needs and ensure a memorable experience through their journey.
AI parenting and its influence in the workplace
In the workplace, there are many ways to leverage AI to make employees’ lives easier. For example, in customer support roles, AI assistants can be implemented to support human staff. This kind of AI can find information related to customers’ questions in real-time, allowing them to complete their tasks faster and more effectively. Additionally, AI brings a host of other benefits, such as being able to generate marketing content, tracking KPIs and even auto-completing emails. AI can also enhance productivity by facilitating meetings and making simple administrative jobs far easier. Need to take minutes in a meeting, or didn’t quite catch what a customer said in a call? AI can summarise meetings and even generate action reports.
However, leaders need to keep in mind that the more AI is used, the more important it becomes to use it responsibly. Organisations must ensure there are strong data collection policies in place, train your team to effectively use AI, and never stop training and parenting the AI systems.
Raising responsible chatbots through AI parenting is a practice that companies should adopt to ensure that the AI is being monitored effectively and cannot interfere with the successful running of a business. Much like your little ones, AI must be nurtured and trained to the highest standard.
