Overcome the barriers to generative AI adoption in the workplace

MIT Open Learning
MIT Open Learning
Published in
7 min readOct 28, 2024
Blue image with three sections about examples, capabilities, and limitations.

By MIT xPRO | The Curve

Artificial Intelligence (AI) is transforming the workplace at an unprecedented pace. But along with this rapid evolution comes a host of challenges that many organizations and employees aren’t fully prepared to tackle.

Generative AI, in particular, is changing how tasks are done, opening up new possibilities while raising questions about its integration and ethical use. In a survey of IT leaders, 86% said they “expect generative AI to play a prominent role at their organizations in the near future.”

MIT xPRO’s new course, Driving Innovation with Generative AI, is designed to help professionals understand the potential of AI and develop the skills to take advantage of its capabilities.

The Differences between Traditional AI and Generative AI

Traditional AI has long been a powerful tool for tasks like prediction and classification. It can do things like analyze X-rays to detect signs of a tumor or use facial recognition to match a person’s face to a database of images. Generative AI pushes the boundaries even further.

By identifying patterns in its training data, generative AI can produce entirely new content — whether that’s generating text, crafting images, or even composing videos. While both types of AI share similar algorithms, generative AI’s true strength lies in its ability to innovate and create rather than merely analyze and predict.

Generative AI has made AI accessible to the broader public in a way that traditional AI never did. As Luke Hobson, Assistant Director, Instructional Design, MIT xPRO, notes, “Generative AI has been the gateway for many people to finally engage with AI.” Language-based tools like ChatGPT and image generators like DALL-E have sparked a wave of creativity, allowing everyday users to interact with AI in practical and imaginative ways.

Hobson points out that chatbots like AOL Instant Messenger’s SmarterChild were early examples of how users could interact with AI — a reminder that AI tools aren’t entirely new. These simple systems, which handled basic tasks like sharing movie times or sports scores, were precursors to modern large language models (LLMs). Fast-forward to today, and LLMs are far more sophisticated, connected to vast data sources, and capable of delivering complex, nuanced responses.

As Antonio Torralba, one of the generative AI course instructors, puts it, “The leap in the last few years has been surprising,” highlighting just how advanced — and accessible — AI has become.

Challenges to Adopting Generative AI in the Workplace

Despite the promises of generative AI, integrating it into the workplace is no simple task. Here’s a closer look at the most pressing hurdles in adopting this technology effectively.

1. Inability to use generative AI to its fullest potential

Many people still don’t know how to maximize AI’s capabilities. In one survey, 64% of executives reported feeling “a high sense of urgency to adopt generative AI,” but over half of those respondents admitted that their organization lacks “the most critical skills.”

One of those skills? Prompt engineering, which involves crafting detailed and precise prompts. Without this foundational skill, many users fall into a common trap: they try using a generative AI tool, generate subpar results, and quickly become frustrated.

Because AI models can produce impressive-sounding answers, it’s easy to assume they’ll work perfectly out of the box. When the output doesn’t meet expectations — often due to a vague or poorly constructed prompt — users may conclude the tool isn’t effective and give up.

Hobson explains that MIT xPRO’s generative AI course teaches prompt engineering. “In Module 3 on large language models, learners go through exercises that cover appropriate levels of detail in prompts, which is a key part of effective use,” he says.

2. Difficulty evaluating and interpreting AI outputs

Writing good prompts is just the start. Generative AI users also need to critically evaluate the outputs they receive. Hobson warns, “These models are designed to make users happy, so they may produce things like fake citations if the information doesn’t exist.”

He points to a well-known case where a lawyer submitted fake case references from ChatGPT in court. While technological improvements have been made, especially with models now connected to the internet, understanding how to interpret the AI’s output remains crucial.

Torralba adds, “If you don’t understand how [a tool] works, it’s easy to misinterpret its output. When you use Google, you get links written by humans. But with generative AI, the responses are machine-generated. The fact that it sounds true doesn’t make it true. You have to know how to interpret the output, and that comes from understanding how the model works.”

3. The rapid pace of change

New tools and capabilities are emerging at an overwhelming pace. Staying on top of the latest developments is crucial, but it can feel like a never-ending task as technology continues to advance.

However, as Torralba notes, “While the technology evolves quickly, that doesn’t mean all jobs are impacted right away.” Many workers have time to learn how to effectively use these tools before the next wave hits, allowing them to gradually integrate AI into their roles.

“The people most impacted by this technology right now are AI researchers,” Torralba adds, emphasizing that it’s those on the front lines of AI development who are feeling the effects of rapid change the most.

4. Unrealistic expectations from management

Many managers approach generative AI with sky-high expectations. But there’s no scenario in which AI can drastically improve productivity overnight. In most cases, it won’t replace human workers anytime soon.

Torralba explains, “It’s not well understood what AI can really do in many jobs. In marketing, for example, AI can assist with the creative process, but it can’t replace humans. If you rely too much on AI-generated content, all companies will end up with the same marketing campaigns. Humans need to be part of the process to extract value from these tools. AI doesn’t do the job for you; it helps you do your job better.”

Hobson echoes this sentiment, comparing AI to early perceptions of Google or Ask Jeeves, when people thought these tools would revolutionize everything. “AI is just another tool to incorporate into your work,” he says, stressing that while it can improve efficiency, it’s not a magic fix.

It’s also important to note that productivity gains from AI aren’t immediate. It takes time to train workers on how to use these tools effectively, and simply introducing AI into a workflow doesn’t guarantee instant results. Workers must first develop the skills to leverage the technology, meaning companies must account for a learning curve before they see significant improvements.

One of the key goals of MIT xPRO’s course is to give workers a realistic understanding of AI’s capabilities and limitations so they can better manage expectations within their organizations.

5. Intensified competition in the job market

The rapid integration of generative AI is transforming the job market by reshaping competition for many roles and putting pressure on workers to adapt.

AI proficiency is quickly becoming a sought-after skill in certain fields. As Luke Hobson points out, “Hiring managers [in fields like instructional design] are increasingly asking about candidates’ knowledge of AI. They’re trying to figure it out for their own organizations.” Being skilled in AI tools can provide a major competitive edge for job applicants.

Ignoring or banning AI isn’t a viable option. “It’s here to stay,” Hobson emphasizes, underscoring the importance of learning to work with this technology rather than avoiding it.

6. Ethical dilemmas

As generative AI becomes more embedded in the workforce, it brings to light potential ethical dilemmas: increased risk of bias, challenges in distinguishing fact from fiction, and even privacy violations.

“AI systems are trained on data, which can raise copyright or privacy concerns depending on the source of the data,” warns Torralba. Privacy issues are especially problematic when employees use unrestricted LLMs, as these systems can unintentionally reveal sensitive information. “Current technology isn’t robust enough to guarantee that won’t happen,” Torralba cautions.

Real-world incidents highlight the severity of these risks. Hobson shares the story of Samsung employees uploading financial records to ChatGPT, resulting in a data leak and a subsequent ban on generative AI. Another case involved Taylor & Francis selling data from academic journals to Microsoft for AI training, sparking backlash. These incidents underscore the ongoing need to navigate ethical concerns carefully as AI technology continues to evolve.

MIT xPRO’s Generative AI Course Addresses These Challenges Head-On

MIT xPRO’s generative AI course is designed to equip learners with the essential skills to navigate the evolving and complex landscape of AI in the workplace.

Rather than simply providing answers, the course focuses on teaching learners how to ask the right questions. As Torralba explains: “The answers will vary by domain and change over time.” This approach ensures that participants can adapt to AI’s rapid advancements and apply the technology thoughtfully in their respective fields.

A key feature of the course is its hands-on, interactive approach. Learners engage directly with AI tools and platforms, gaining practical experience in real-world scenarios. The course covers key skills such as prompt engineering, helping learners craft detailed and precise prompts to maximize AI’s potential. It also emphasizes the importance of critical thinking, teaching participants how to assess AI outputs — whether it’s spotting bias or fact-checking generated content.

Ethical concerns, from privacy issues to data handling, are thoroughly explored, giving learners the knowledge to responsibly navigate the risks that come with AI. By focusing on the capabilities and limitations of AI, the course ensures that participants can manage expectations and make informed decisions as they integrate AI into their work.

Ultimately, the course provides a holistic framework, blending tangible technical skills with a deeper understanding of AI’s ethical and practical implications to prepare learners for the future of AI in the workforce.

To learn more about what the generative AI course has to offer, check out the course page.

Part of MIT Open Learning, MIT xPRO provides professional development opportunities to a global audience via online courses and blended programs.

--

--

MIT Open Learning
MIT Open Learning

Published in MIT Open Learning

News, ideas, and thought leadership on the future of learning. Discover more at openlearning.mit.edu.

MIT Open Learning
MIT Open Learning

Written by MIT Open Learning

Transforming teaching and learning at MIT and around the globe through the innovative use of digital technologies.