As collaboration with AI expands, so does workers’ optimism

Just a month before ChatGPT made its debut in November 2022, veteran AI researcher Sam Ransbotham published one of the first in-depth surveys about AI’s emerging influence on workers. In many ways, it predicted how things would unfold as generative AI found its way into the hands of millions.

His report—a joint effort between Boston Consulting Group and MIT Sloan Management Review—presciently noted that the use of AI on the job boosted individuals’ sense of competence, autonomy, and connectedness to their teams. A 2024 Freshworks study in 2024 later showed how AI bolstered employees’ sense of purpose, not just their productivity.

Ransbotham, an author and professor of business analytics at Boston College’s Carroll School of Management, has since seen gen AI and now agentic AI become part of the fabric of daily work.

We interviewed Ransbotham recently to get his latest insights on where AI’s influence on people and work is headed. Will AI continue fostering individual and organizational value, or could more autonomous forms of AI possibly derail things? Following are highlights of our conversation. 

In 2022, your research showed that employees’ use of AI was translating to organizational value. Has this continued with the rollout of newer AI tools?

I think it’s net positive, but it’s not clearly a no-brainer that it works only in that direction. There are mechanisms that could go both ways. Self-reliance, self-determination, competency, autonomy, relatedness—these could potentially be undermined if using AI tools makes workers feel less competent, decreases autonomy, or reduces human connection.

But that’s not what we’re seeing on balance. People are quickly figuring out the strengths and weaknesses of these tools, and getting excited about the capabilities while recognizing the limitations.

How is gen AI, in particular, helping workers get more value on the job?

Consider platforms like Slack where overwhelming information flow can consume your entire day. Generative AI’s summative function—processing content and highlighting what’s important—is making people more confident and autonomous.

Rather than asking other people what was discussed or having to spend time figuring it out, they can use the tool to quickly get up to speed. For example, a salesperson dealing with numerous clients can summarize the latest developments of a deal without spending hours sifting. 

Before gen AI and copilots, your 2022 study suggested that workers were beginning to see AI as a coworker. Does this idea still hold true?

Based on preliminary data from our current survey [which is now underway], this sentiment is increasing. When we initially included that question, it was a last-minute addition—we had extra space in the survey and thought, “Let’s see what people think.”

We were surprised by the positive response because it contradicted the prevalent media narrative that AI would threaten people’s jobs. Our focus was on knowledge workers and managers. Other segments of the workforce might have felt, and still might feel, differently. A graphic artist, for instance, might not view AI as a coworker.

Where are companies using AI copilots in particularly effective ways?

Customer service centers are among the most effective examples because they use fast feedback. Experiential learning involves concrete experience and immediate feedback. It doesn’t help if support agents get feedback on calls three months later, but it’s tremendously valuable when there’s real-time guidance.

One example is Estée Lauder. Makeup is inherently personal, and they’ve equipped their in-store sales associates with AI tools that provide real-time information and visualization. Associates can take a customer’s picture, instantly show different makeup options, and adjust based on preferences—a “try on 30 lipsticks in 30 seconds” capability.

This creates a learning partnership where both the sales associate and AI learn from each other. The associate provides feedback to the AI, and the AI suggests options the associate might not have considered. When we think about AI copilots, we tend to think about pilot and copilot, but this model is more like paired copilots, each helping the other one.

Your research suggests that workers’ optimism about AI improving job performance is on the rise. How do you account for that?

When we first asked about AI’s impact on jobs in 2017, it was too abstract for most respondents. AI was something they read about in the news, not something they personally experienced. By our 2024 survey, the number of workers hopeful that AI would improve their ability to do their jobs had risen from 70% to 84% while the number fearful dropped from 31% to 20%. 

I was surprised that hope was increasing while fear was decreasing, because media coverage tends to emphasize job displacement. I feel like every news outlet that contacts me wants to talk about that story—about machines taking jobs—and here was compelling counter-evidence of that. 

That growing optimism reflects employees’ direct experience with these tools. People now understand both what AI can do, such as writing first drafts, and what it can’t do. The tools are still young, and people are quickly figuring out their strengths and weaknesses. And they’re excited about their strengths.