Click to play background video

Click to pause background video

Everywhere you look, someone is talking about AI. And you're expected to make decisions about it. But what is it? And more importantly, what should you consider when deciding which AI product to invest in? Read on to learn how you can choose wisely.

 

Here you’ll learn:

Standard AI structure ↓

Good vs Great AI ↓

AI at LinkedIn ↓

Resources ↓

In October 2024, our CEO Ryan Roslansky shared some exciting news to unveil LinkedIn's first agent, Hiring Assistant. But he also shared our how we’ve brought the best of LinkedIn together to create this impactful AI tool. At a high-level, our tech stack is:

Is the technology and infrastructure powerful enough to handle my needs?

Is the data updated regularly, and relevant to the talent industry?

Is it built responsibly and will it be compliant in the long-term?

 

Keep up with today’s thought leaders:

Ryan Roslansky, CEO of LinkedIn

Hari Srinivasan, VP of Product at LinkedIn

Erran Berger, VP of Product Engineering at LinkedIn

Blake Lawit, SVP and General Counsel at LinkedIn


Click to play background video

Click to pause background video

Adopting AI inside your organization can be a big change. Beyond the tools you roll out, the way you roll them out makes all the difference. That’s where change management comes in.

But just as AI requires us to rethink how we work, it also requires us to rethink how we approach change management.

 

Here you’ll learn:

Change Management 101 ↓

Change Management for AI ↓

Resources ↓

 

All with pro tips and examples from our change management team at LinkedIn. Ready to get started?




When it comes to AI, change management is less about setting hard deadlines and more about planning in short bursts so you can learn and adapt, in real time.

Allocate 3-6 months to get leaders on board and engage your broader team. Beyond that, take a flexible approach so you can incorporate new learnings along the way.

How LinkedIn is doing it:
We’re at the forefront of providing innovative AI products and features to our customers, and we’re taking the same care in how we bring AI to our employees. Instead of driving towards launch and then transitioning to business-as-usual, we’re taking a scaled approach to help employees understand how AI impacts their day-to-day work. That means enabling learning, application, and adaptability from the start so employees can evolve their understanding of AI on an ongoing basis.

Example:
We designed our GenAI Upskilling program to give employees a foundational understanding of GenAI tools – and keep us innovative and future-ready as an organization. Since AI is constantly evolving, we realized that the program would need to do the same. By grounding ourselves in a clear goal, we were able to divide our work into phases. We started with a small group of new users, which helped us identify ways to boost the use of GenAI at scale. Most importantly, it uncovered the need for personalized change efforts down the road, unlocking longer-term planning for a sustainment team with AI champions. Today, our champions stay on top of new AI developments and apply learnings to their teams and functions, with a centralized body to maintain a consistent experience across our organization.

In change management, your change narrative is your first opportunity to build buy-in. When it comes to AI, it’s also your moment to address people’s anxieties by communicating the value AI will provide.

Highlight how it will support people vs replace them, and tie it back to larger goals like automating operational tasks to create more face time with candidates and clients.

Pro tip: Use AI to talk about AI. Copilot can help!

How LinkedIn is doing it:
At LinkedIn, a clear change narrative is integral to the way we introduce GenAI tools. By framing our ‘why’ in a human way, we help employees understand how AI can enhance their capabilities and help them use their skills for higher-level problem solving, creativity, and innovation.

Example:
We launched our Coaching for All program to provide one-on-one coaching to employees. Typically, our narrative would have focused on the benefits of the program, like meeting with a certified coach to discuss personalized goals. But this program was designed to empower employees in the age of AI, so we needed to shift our approach. By focusing on the larger why, we connected the program to our long-term vision and built buy-in from the start. Our final why statement: ‘With new tools, technology and ways of working emerging every day (hello, AI), it’s critical for everyone to continuously upskill. And for our company to stay ahead, we need to ensure that all of us have the best training and coaching.’ 

AI is constantly evolving, so learning happens in short and intentional bursts. Start by getting clear on what your people know and don’t know.

Even if they’re already using AI, they might need to brush up on hard skills (like prompts) and soft skills (like relationship-building) to prepare for what the future of work will look like. Once your team has a foundational understanding, create a system that lets them upskill as new updates, features, and best practices emerge.

Pro tip: Be transparent about new developments, new skill gaps, and how you’re planning to fill them.

How LinkedIn is doing it:
When it comes to AI, we continue see the value of ongoing learning. Our traditional change management process follows a steady learning curve with a clear definition of success, but AI-driven change is a moving target that requires our employees to learn, apply, and adapt through short but intentional bursts. 

Example:
We created our GenAI Upskilling program to help employees build a foundational understanding of GenAI. Our goal? To give them the skills to innovate with AI in hand and keep our organization ahead of the curve. We quickly realized that the program needed to evolve in line with AI, so we decided to take a phased approach. Our team started with basic training to get employees comfortable with GenAI, using examples of how different tools improve their day-to-day workflows. From there, we identified the need for persona-based, hands-on workshops to familiarize employees with prompt building, bots, apps, and agents, along with experienced SMEs to live-demo their insights. This drove curiosity and engagement along the way, and we continue to see the results through the sustained usage of AI tools.

Change management typically starts with your executive team. With AI, it starts by engaging leaders across your organization so you can assess and manage the impact, across the board.

Bring your leaders in early. Make sure they understand the tool and its value so they can build buy-in within their teams and help activate your change champion network at scale (see Section 2: Work from the inside out).

Pro tip: Hold shared briefings so leaders can ask questions and flag barriers, ahead of time.

How LinkedIn is doing it:
At LinkedIn, we’re seeing AI create room for more strategic and meaningful work. Our employees are less siloed and more open for human collaboration. AI-driven change is moving beyond individual teams and tasks with implications on global systems, workflows, and processes. Our change management approach isn’t just based on the immediate impact of AI, but on the broader consequences of an interconnected system. In line with that, our change impact assessment – designed to measure the potential impact of change on key stakeholder groups – has evolved to capture global shifts in workflows, dependencies, and processes.

Example:
We experienced this shift in real time when we redesigned our new-hire onboarding process. Onboarding typically includes a set of local processes, like training on specific tools and access to internal systems. With AI, we knew we had to consider a broader set of variables that come with being a global and interconnected organization. A thorough impact assessment helped us anticipate and accommodate global needs, like ensuring access to systems across multiple time zones, managing security protocols for remote employees, and tailoring our chat bot and training materials to different cultures and legal requirements. 

Dig into our approach

Discover the finer details

Expert Contributors: Our Change Management team at LinkedIn

Drushti Gandhi, Director, HR Change Management Lead at LinkedIn

Roberta Chew, Head of HR Enablement at LinkedIn

Christine Nguyen, HR Senior Change Manager at LinkedIn

Stacy Zhong, HR Senior Change Manager at LinkedIn

Kristen Lahoda, Senior Manager, HR Change Management at LinkedIn

AI will continue to evolve, and so will change management for AI. Watch this space as we add new developments, recommendations, and resources.


Click to play background video

Click to pause background video

How can you trust your AI tools to get it right?


Today, every AI product on the market is promising to revolutionize your work, right out of the box. But how do you know if what you’re using will actually deliver results that help your teams? That’s where accuracy comes in. And understanding it can make the difference between AI that delivers and AI that disappoints.
Image of a woman wearing glasses

Start with strong inputs
Quality results start with quality inputs. This is a responsibility you and your provider share.

Run it through the right model
AI models get better every day. You need to ensure that your providers are working with a model that’s best suited to the task at hand.

Add oversight
The goal of AI is to present you with the best options, so you can make the best decisions. But no matter how good it gets, some oversight is still necessary.

Illustration representing what goes into AI accuracy
A person drawing on a white board
Two chat bubbles icon

You get out what you put in

When it comes to accuracy, that's an understatement. Generating high-quality candidate recommendations and job descriptions starts with high-quality data and prompts. But don’t worry, that’s not all on you. It’s a shared responsibility between you and your provider.

Your provider’s role

Your provider is responsible for making sure their AI is trained on high-quality data that’s current, relevant, and fair.
This requires asking a few key questions.

Icon of a map pin

Your role

You are responsible for crafting high-quality prompts, giving feedback along the way, and helping others on your team do the same.

Chat icon
Illustration of a chat bubble
A person working on a computer
Illustration representing an AI model

Would you use a flip phone in a world of smartphones?

Sure, you could perform basic functions, but you'd miss out on the speed, capabilities, and up-to-date features of a more current option. That’s what using the wrong AI model is like. And because they’re constantly improving, it’s crucial to choose a provider who prioritizes using the most appropriate model for each specific task and who's committed to regularly reassessing their approach to incorporate the latest advancements.

Look for providers that are:

A person working on a computer
Illustration representing an AI model

Even the best AI needs an extra set of eyes

No matter how good AI gets, some oversight is always needed to make sure the results you’re getting are up to your standards. This is especially true as your AI tools get used to your unique needs and working style. Treat AI output as a starting point. By actively engaging as an editor and guide, you help it learn and adapt, ultimately leading to more accurate and relevant results.

Illustration representing verifying source data

Verify against source data
While AI can process large amounts of information, it's crucial to cross-check its findings with the original data sources (e.g., resumes, performance reviews). This verification step helps confirm the accuracy and completeness of the AI's analysis and ensures no critical information was overlooked or misinterpreted.

Illustration representing AI seeking broad perspectives

Seek broad perspectives
AI recommendations should be reviewed by a cross-functional group of individuals within your talent team (recruiters, hiring managers, HR business partners). Different perspectives can help identify potential blind spots or unintended consequences in the AI's output, leading to more balanced and well-rounded decisions.

Illustration representing looking out for bias

Be on the lookout for bias
Ideally your provider will work to ensure against bias, but it’s still a good idea to continuously monitor AI outputs for unfair or skewed results related to protected characteristics (e.g., gender, ethnicity). Implement processes to identify the sources of bias (in training data or model design) and take corrective actions to reduce or eliminate them.

A team working together

We’re bringing our best to the table to build AI you can trust.

More dynamic data
Our tools are trained on up-to-the-minute talent data from our Economic Graph, with 5M profile updates per minute.


Illustration of a rising bar graph

Cutting-edge models
As part of Microsoft, we have access to the latest and greatest technologies.


Illustration representing cutting-edge models

Constant iteration
Our teams are constantly testing new models as they emerge to ensure that our products are equipped to deliver the most accurate results.


Illustration representing product iterations
A person using the LinkedIn mobile app

While a true accuracy process takes a while to refine and get right, there are a few things you can do today to make sure your team is getting the most accurate results from their tools, and developing a trusted relationship with their AI tools.