Will You Be More Receptive to AI-Generated Feedback?

Let's see what research suggests. The answer may surprise you. Or it might not.

For as long as organizations have existed, feedback has too.

Actually, that might be an overstatement.

Still, feedback in the workplace has a long-standing history and has consistently been used as a tool for growth and transformation.

As you might expect, a company that knows how to leverage feedback effectively typically gains an edge over its competitors. This edge often takes the form of increased profits, a larger market share, among other long-term advantages.

Every organization aspires to attain this advantage, but only a few manage to truly harness the power of feedback effectively to bring about meaningful change.

It is no secret that artificial intelligence has already begun to permeate a wide range of corporate functions. Even if not leading the way, AI is present in almost every organization in some capacity.

Hence, it comes as no surprise that companies are contemplating (or have already begun) using AI to enhance the way feedback is delivered.

The key question here, however, is not just how accurate or unbiased AI-powered software is when delivering performance insights. It’s also about how employees perceive this kind of feedback.

A recent study published in Nature, a premier global science journal, goes into detail about AI’s potential in provide feedback to employees. It compares the impact of AI-generated positive feedback that focused on strengths and accomplishments, with negative feedback that highlights areas for growth and opportunities of advancements.

The study also considers another important factor: the role of social support from managers and teammates, and how that affects an employee’s perception of their performance and their belief in their own abilities (self-efficacy).

I’ll get into the details of each factor soon, but let’s first talk about why this research matters.

AI for Feedback? Why exactly?

Since we cannot stop artificial intelligence from becoming part of our workplaces, it is crucial to ensure that humans and AI form a symbiotic relationship rather than an exploitative one.

As long as people—actual talent—are a part of the workplace, nurturing this talent will remain a point of concern and contention. Therefore, using AI and defining its role in supporting talent development within organizations is clearly an area worth studying.

A growing number of studies indicate that artificial intelligence can significantly support employee development by using data-driven methods to assess key metrics such as turnover risk and learning and development needs. Its applications are wide-ranging and relevant at every level, from individual employees to entire organizations.

When we consider organizational coaching, for instance, several studies now suggest that AI-led coaching can be just as effective as human-led coaching. What makes AI-based coaching even more promising is how accessible it is to employees.

Moreover, employees don’t have to fear judgment or negative evaluations, as AI doesn’t care if they make mistakes. And unlike traditional coaching, there’s no need to coordinate busy calendars. AI-driven coaching is available on demand.

Given the proven effectiveness of AI in the domain of learning and development, it’s no surprise that researchers are now exploring its potential to deliver effective employee feedback. This interest is also driven by the reality that many managers lack the time, or in some cases the necessary skills, to provide meaningful, constructive feedback.

In such situations, AI can serve as a valuable assistant to the manager, benefiting both the manager and the employee, while also strengthening their overall relationship.

The point of the AI revolution at work, as I see it, is that employees don't need to take on an overwhelming load of tasks and try to handle everything on their own. It also doesn’t mean that humans should be entirely replaced by machines. It’s about creating a synergistic relationship, and workplace feedback is just one of many areas where that relationship can be established.

Furthermore, as AI becomes more capable of holding human-like conversations, its potential to deliver thoughtful, personalized feedback continues to grow.

Now, back to the study.

If we take a look at the results, a number of interesting findings emerge.

Firstly, at the individual level, feedback usually relates to recent behaviour, performance, and immediate goals. However, AI can go beyond surface-level feedback by helping create a detailed roadmap that also highlights long-term objectives.

Secondly, it is worth noting that when the AI software delivers positive feedback, employees feel confident in their abilities. In scientific terms, positive feedback enhances self-efficacy.

And you know what’s fascinating? This principle applies to human-to-human interaction as well. When you receive positive feedback from your manager, it tends to have a greater impact than negative or even 'constructive' feedback.

Now that we are progressively moving toward an AI-integrated workplace, it has been found that regardless of the source, positive feedback is the key differentiator.

But we often receive negative feedback too, so let’s take a moment to look at its implications.

When employees receive negative feedback and feel unsupported by their company, they tend to disregard the feedback. In research terms, it has no significant impact.

However, when employees feel that their manager and colleagues genuinely care about them and they are well supported, the findings shift dramatically and negative feedback becomes just as effective as positive feedback.

Are these findings really that surprising, though? I mean it makes perfect sense that people care more about their work when they feel that the people around them care about them.

What’s truly remarkable, however, is that these findings hold true even when the feedback comes from AI. Positive feedback is effective whether it’s delivered by a human or by the AI that may have replaced them. And negative feedback still has an impact but only if employees feel supported, even when it comes from AI.

Naturally, the support must come from leaders and peers, not from the AI software delivering the feedback (in case that wasn’t obvious).

Why is this the case? (theory time!)

According to Cohen and Wills’ buffering hypothesis, emotional support protects individuals from the harmful effects of stressors like criticism or failure. It serves as a psychological shield, reducing the emotional toll of difficult experiences like negative feedback.

Another curious finding, contrary to the widespread suspicion surrounding AI, is that when AI delivers negative feedback, people are more likely to trust it. The reasons for this are still debated. One possibility is that AI is seen as less likely to be biased toward any individual. It may also be viewed as a more credible source, given its training on large volumes of data. What is important is that this kind of trust makes emotional support in the workplace even more important.

However, when it comes to the workplace, not all forms of support are equally effective and the type of support offered also changes how feedback is perceived.

In this study, when employees received instrumental support or tangible help, for example a colleague taking over part of their work, it caused them to question their own abilities, thereby decreasing self-efficacy.

The same went for informational support, which includes advice, guidance, and technical input. Even though it was well-intentioned, it left employees feeling inadequate, dependent, and overall worse about themselves.

Only emotional support, that is genuine care, encouragement, and empathy, was found to consistently improve employees’ confidence and outlook.

These findings are eyebrow-raising to some extent, but they also feel relatable. I mean, if my manager took away a responsibility I had been entrusted with, I’d probably wonder if I had failed in some way, too.

That said, it's important to note that this particular study was conducted in Japan. Cultural differences must be taken into account when considering how universally applicable (generalizable) these findings are.

What I’m really curious about, though, is how would you respond to AI-generated feedback? Would you be more open to it? Would you trust it more (or less) than feedback from a manager?

Let me know your thoughts in the comments!

Until next time,

Enjoyed the read? Consider subscribing!