Jared Spool 指出 AI 应用存在双方用户体验问题

Recently, Satya Nadella, the CEO of Microsoft, admitted to a Bloomberg reporter that he uses Microsoft’s AI CoPilot to summarize emails his team sends him, so that he doesn’t have to read the messages or attachments.

Imagine being someone who worked for hours writing something for Nadella, only to find out the CEO outsourced reading it to an algorithm? How would you feel?

People don’t like being on the receiving end of AI usage.

Job applicants hate it when they learn that a company they’ve applied to has used AI to screen their application, especially if it then rejects them.

At the same time, hiring managers don’t like it when they discover a job candidate used AI to embellish their résumé and other application materials.

College professors dislike it when their students use AI to do their assignments, while college students complain when they learn their professors are using AI to create their lectures.

And nobody likes dealing with a customer service chatbot. For many people, their first reaction is to try to convince the chatbot to refer them to a real human. This behavior has become so prevalent that big brands, like United Airlines, have begun advertising human-only customer service chats. Bypassing AI is now a marketing advantage.

All this is to say that if you’re implementing AI functionality, you’ve got a two-party UX problem. While Microsoft’s Nadella is proud of his AI summaries, it’s unlikely the people who work hard on their emails to him are too happy about the fact that he never reads them. If they knew he only wanted a summary, why didn’t they just write him a summary instead of letting the large language model hallucinate their way through summarizing it?

There are two parties in these scenarios: The people using AI to summarize the emails and the people writing them in the first place. The job candidates and the hiring managers. The college students and the professors.

When we’re working on the user experience of AI functionality, we need to think about both parties. This consideration is new to our UX practice.

In the past, if we were working on the UX of an email client or an application management system, we’d focus only on one party. We’d focus on the email recipient or sender, but not both. We’d focus on the job applicant or the application reviewer, not both.

But many AI transactions involve multiple parties, and now we need to think about all of them simultaneously. We want to find a way to make both sides feel that our AI functionality improves their lives, not worsens them. 

AI opens us up to social problems, where there are multiple sides to any design. The trick is to identify solutions that improve everyone’s lives, not just a single party’s experience at the expense of everyone else.

When you’re designing AI functionality, how are you focusing on the needs of all the parties involved, not just your primary audience?

The post Nobody Likes Being on the Receiving End of AI appeared first on UX Articles by Center Centre.