Editor’s note: Besides following tech developments and innovation, our author is a musical composer (Juilliard-trained). He has provided a musical composition for you to listen to while reading this column. This piece is called “Dialog Matters.”
Many people feel a shift in how relationships and trust are built. Ratings, reviews, and metrics have increasingly replaced personal connection or nuanced evaluation in interacting with professionals, companies, and leaders. This emphasis on numbers can leave us feeling reduced to a data point rather than valued as individuals. As systems grow more data-driven, the human element risks being sidelined. People are starting to push back, seeking relationships and services where they feel genuinely seen and valued. For doctors, the focus on patient satisfaction surveys might sometimes overshadow genuine care. For politicians, public opinion polls often shape decisions more than long-term strategies. Many companies prioritize ratings and metrics like customer retention or net promoter scores, which can make interactions feel transactional instead of meaningful. One way people’s real feelings are often dismissed is through polling, with its inherent confirmation bias.
Confirmation bias is the tendency to seek, interpret, or remember information that confirms one’s preexisting beliefs or opinions while ignoring or dismissing evidence that contradicts them. Bias confirmation often plays a role in polling, especially when the questions are designed or framed in a way that reinforces certain assumptions or beliefs. In politics, for example, polls can be conducted with leading questions or in ways that attract responses from people who already hold a certain viewpoint. This can create the illusion of widespread agreement on an issue, even though it may not represent the true diversity of opinions. Additionally, media outlets or political groups might selectively report polling results that align with their narrative, further contributing to confirmation bias. Instead of truly exploring people’s thoughts, the process reinforces existing beliefs and preferences.

AI has great potential to improve interviews and decision-making by reducing bias confirmation, provided it’s implemented thoughtfully. AI could use natural language processing (NLP) to ask dynamic, follow-up questions that probe deeper into relevant topics without leading the interviewee. This helps gather richer data while avoiding the biased assumptions seemingly validated by multiple-choice or yes/no answers. AI bots don’t get tired, frustrated, or defensive like human agents might, reducing the risk of miscommunication rooted in bias.
This can permit a much-improved interaction, different from the scripted algorithmic responses to which people are increasingly subjected by humans and unintelligent machines. And AI systems can analyze responses more objectively, emphasizing the substance of answers uninfluenced by personal charisma, body language, or irrelevant traits. This allows decisions to focus on qualifications, ideas, and competencies.
When interacting with mindless processes, automated or not, I feel my time is being wasted and I am being disrespected. Sometimes, although I may clearly see both the problem and a potential solution, I may be subjected to a ridiculous script. Here’s a recent example: new callers to my cell phone are being sent directly to voice mail, without my phone ever ringing. I have spent considerable time with one of the major telecom companies trying to address this issue. The phone company attempted to call me ten times, each time being sent to voicemail. That’s right—even the phone company couldn’t call me. Eventually, after spending many hours over multiple days with online scripted chats and phone calls, I got to a live person. I explained my understanding of the situation, and he fixed it in ten seconds.
I have had many conversations with AI; it never once wasted my time or disrespected me. I asked an AI how to break through the phone company’s system to get to a real person, and it told me explicitly how to do it. I followed its directions, and the lingering, frustrating, time-wasting problem was instantly solved. This alerted me to AI’s potential to improve these types of interactions.

Here is another example of how I use AI, which saves me an incredible amount of time, effort, and frustration. As a composer, I use a scoring notation program called Sibelius. It has been around since 1993 and is now the most widely used application for notation. Unfortunately, over the years, several changes of ownership have degraded the customer service experience. The instruction manual has grown from 140 to 762 pages long. The user interface has changed enough times to confuse all the professors who use it to teach. Almost every member of the development team that created this powerful and complex software has left. The YouTube tutorials and the online forums contradict each other. Communicating with the company is unforgivably frustrating.
I tried an experiment. I asked ChatGPT to read the entire 762-page manual and then asked it how to accomplish specific tasks. It instantly provided clear instructions on how to proceed, and when I found contradictions, I asked it, “Are you sure about that?” Then it went off and read all of the forums and tutorials, searched for videos, provided references, and got me on track in minutes instead of hours. Furthermore, I do not have to type in these questions but can speak them, and ChatGPT can speak back to me while I am using the app, so it is like having a consultant with me who is far more knowledgeable than any human.
I have since tried this experiment with other powerful, complicated software, as I seem to have multiple apps with hundreds-page manuals. ChatGPT listens to what I am requesting, answers the questions, and even remembers the context within which I am asking them, which makes me feel like it knows who I am far more than any of the companies I try to get answers from.
I think this will result in people replacing Google as their search engine with AI large language models (LLM). If Google does not respond, it will suffer greatly financially as the value of what it delivers in a Google search is far less than what an AI LLM delivers.
The next time you try to use even Word, Excel, or PowerPoint, all of which have gigantic manuals, and have something you want to know how to do, ask an AI. You will be surprised at how much faster this is than either asking Google or trying to read the manual yourself.