Prediction: AI Will Rescue Performance Management
What's in this month's Newsletter?
Main Feature: Prediction: AI Will Rescue Performance Management
(Significant!) Personal and Professional Updates
Prediction: AI Will Rescue Performance Management
AI-Powered Performance Management: Transforming Employee Growth and Development
You know what probably isn’t all that great at your organization? Performance Management. I’m not sure I know anybody that thinks their organization does Performance Management really well. It’s like swimming the butterfly stroke: You can do it, sort of, maybe for a short period of time, but you’re probably making a lot of small mistakes that add up to bad form. To sustain it at a high-level is exhausting.
Managers struggle to gather feedback from multiple sources, compile data, and craft narratives to justify ratings. Performance is hard to measure and prone to cognitive biases. Rating systems are too rigid. Employees, on the other hand, face anxiety as they await the verdict on their performance. The process is far too slow. The result is a time-consuming, stressful exercise that fails to foster a true culture of growth and development.
Here’s some good news (well, good if you don’t fear change): improving Performance Management is low-hanging fruit for AI. Harnessing the power of AI language models, like ChatGPT, will revolutionize the way we assess and improve employee performance, creating a more engaging and effective experience. As the most progressive organizations start to incorporate AI language models into their Performance Management processes beginning this year, here are some predictions (suggestions!?) for AI-enhanced Performance Management:
AI-Driven Performance Management
Real-Time Feedback: Imagine an employee, Mark, has just completed a presentation and his boss, Jane, wants to give him feedback on his performance. Instead of waiting for a formal conversation or trying to find the right words to convey her thoughts, here’s how Jane could use a chat platform (e.g., Slack, Microsoft Teams) integrated with an AI language model like ChatGPT. First, Jane fires off an unvarnished message to the AI model summarizing her thoughts: "very informative, poor pacing, too many slides." The AI model processes Jane’s input and generates a well-crafted, constructive feedback message in language tailored specifically for Mark because it has trained itself on his communication preferences. Jane then reviews the AI-generated message, quickly makes necessary adjustments, and sends it to Mark right away, allowing him to apply the insights to his next presentation.
AI language models will help facilitate these instant feedback exchanges. Removing the formality and effort typically associated with giving feedback helps employees feel more at ease sharing their thoughts, resulting in a more comprehensive picture of each employee's performance. Instead of waiting for formal check-ins or performance reviews, employees and managers can quickly and routinely discuss performance, fostering a culture of continuous growth and open communication.
Data Aggregation: The AI language model collects and analyzes data from multiple sources, such as emails, project updates, and peer feedback, providing managers with a comprehensive view of each employee's performance. This allows managers to recognize top performers, address workload imbalances, and make more informed decisions. Think of it as a super-360 that is analyzing far more than formal review-time feedback to create a more holistic view of performance.
Streamlined Goal Setting: AI language models assist in setting SMART (Specific, Measurable, Attainable, Relevant, Time-bound) goals, ensuring they are well-defined, realistic, and aligned with the organization's objectives. Employees and managers can collaborate on goal-setting, creating a shared sense of ownership and accountability, and use the AI to compare these goals to others across the organization.
Bias Reduction: Let's say an experienced manager, Kate, is evaluating her team members' performance over the past year. She starts drafting feedback for one of her newer employees, Thomas, who recently joined the team out of college. In reviewing Kate’s evaluation data over a period of many years and statistically comparing her evaluations to those of her peers, the AI model finds that Kate may be affected by an age or tenure bias, as she tends to review newer, younger employees more critically. The model then alerts Kate of these patterns and prompts her to rethink her review as well as suggests alternative, unbiased language.
An AI model can detect patterns indicative of biased feedback, such as gender, age, or tenure-based biases. Highlighting these patterns for managers can help them reassess their evaluations and make fairer, more objective decisions about employee performance. It will also be able to scan performance data across time and space without recency bias, halo and horn effects, and more. AI will make reviewing performance fairer.
Customized Development Plans: The AI language model crafts personalized growth plans for employees, focusing on their unique strengths and areas for improvement. These tailored plans help employees stay engaged in their development and ensure they receive the support they need to succeed. These customized plans will generate experiential learning activities and recommendations for learning resources internal and external to the organization’s learning management systems.
Predictive Analytics: AI will help organizations identify employees at risk of underperformance or burnout by analyzing performance trends and flagging potential concerns. This allows managers to proactively address issues and provide support, ensuring employees remain engaged and productive. The recommended types of support can be tailored to fit each employee’s schedule, learning style, communication preferences, and more.
Enhanced Collaboration: Suppose you’re trying to put together a team for a new project. You could turn to your AI model, which is already well-versed in all of your employees' skills, experiences, tendencies, current workload, etc. The AI quickly identifies the ideal team of available team members whose unique strengths would complement each other and contribute to the project's success. It could craft a custom message to each person and their supervisor asking for their participation, and then scan calendars to automatically find times for everyone to meet (with a suggested meeting agenda built right in).
AI language models can facilitate cross-functional collaboration by identifying complementary skill sets and suggesting projects that benefit from diverse perspectives. This promotes a culture of innovation and helps employees develop new skills by working alongside colleagues from different backgrounds.
Training and Learning Optimization: The AI model analyzes employee performance data to recommend targeted training and learning resources for the organization which is based on the cumulative discovery of each individual's needs and interests. It will be able to recommend specific training vendors, learning opportunities, and even create learning and training content from scratch. This maximizes the impact of professional development efforts and ensures that employees are continuously learning and growing.
The Future of Performance Management: Embracing the AI Revolution
Depending on your organization’s culture, incorporating AI into Performance Management processes may be met with initial resistance. You may have to incorporate a little Dwayne “The Rock” Johnson change management. But the potential rewards are immense and the flood of AI into Performance Management can only be contained for so long. By embracing AI-driven solutions, organizations can create a more efficient, equitable, and supportive environment for employee growth.
Have you already encountered AI-driven Performance Management in action? Can you think of any other ways AI language models could reshape the workplace? Especially if you’re reading this and you have a say in your organization’s Performance Management processes and want to partner and brainstorm how to be on the leading edge of AI-powered Performance Management, let's connect and continue the conversation.
_____
Note: I used the new ChatGPT-4 to help construct the section above. For fun, I wanted to see whether it wanted credit for helping me, so I asked it: “Given that this is a newsletter appearing on my Substack (and not a formal article, or something), ought I to mention that you helped me draft this piece?”
It replied: “As a newsletter on your Substack, it ultimately depends on your preference and the level of transparency you'd like to maintain with your readers. Mentioning that an AI language model, like ChatGPT, assisted you in drafting the piece could provide an interesting talking point and showcase a real-world application of the technology. On the other hand, you may choose to keep the focus on the content and ideas themselves without disclosing your collaboration with an AI. There's no strict requirement, so it's up to you and how you believe your readers would perceive and respond to this information.”
So there you have it :)
Personal & Professional Updates
Some big personal news on two fronts: I am excited to share that Jessie and I are expecting another baby girl in early October! Charlotte is extremely excited to be a big sister and we are rather relieved that it will be a baby girl… as Charlotte has been forcefully insisting that she have a baby sister for many months now.
If that weren’t enough of a change, we’re also planning on moving up to New York! We’re actively on the hunt for a home in Westchester and hope to move in before the baby is born. The impetus for the move is to be closer to family so that Charlotte and Baby #2 can spend more time with their grandparents. Leaving our friends and community around DC will be very difficult, though, and this pending move doesn’t feel fun even if it feels smart.
(So if you know of anyone selling a home in Westchester, let me know….).On the professional side of things, I have been deeply engrossed in exploring the latest in AI. I have only an 8th grade understanding of how the latest technology works on the inside but have been trying to become an expert on how to apply it effectively. So expect to see quite a bit more from me about – and created by (!) – AI in the coming year.