Learn how to successfully navigate AI projects with five critical steps: selection, development, evaluation, adoption, and management. Understand the challenges of trust and user adoption in AI, balancing speed and effectiveness, recognizing and changing course in AI experiments, and the cyclical nature and value of AI projects.
Choosing the right project is critical for AI success, focusing on impact and feasibility while understanding the intended audience helps involve them in the process.
Evaluation is crucial for AI projects, requiring attention to ethical considerations, unintended consequences, long-term effects, and constant evaluation and improvement.
Deep dives
Failure Rate of AI Projects
AI projects have a higher failure rate compared to traditional IT projects. This is due to the probabilistic nature of AI, which adds complexity and uncertainty. AI projects can fail at different stages, such as selecting a project that does not add value, low accuracy of algorithms, or the presence of biases. Trust is crucial for AI projects, as users need to believe in the product and its developers.
Selecting the Right Project
Choosing the right project is critical for AI success. The impact and feasibility of the project should be considered. Impact focuses on aligning the project with the organization's strategic goals, while feasibility examines data availability, infrastructure, and ethical implications. Misalignment between impactful projects and data scientists' preferences can occur. Understanding the intended audience (internal or external facing projects) helps involve them in the process.
Effective Evaluation and Continuous Improvement
Evaluation is crucial to ensure AI projects are effective and meet their objectives. Due to AI's complex interaction with the ecosystem, evaluating key metrics alone is not sufficient. Ethical considerations, unintended consequences, and long-term effects should be audited. Experimentation plays a vital role in improving algorithms, and companies should focus on rapid but safe testing. Paying attention to users, maintaining trust, and testing initial hypotheses are essential for success. The iterative nature of AI projects requires constant evaluation and improvement.
Unfortunately, you can’t set up your organization’s artificial intelligence projects like just any other IT project. By their nature, AI endeavors are quite different and suffer high failure rates. But there are proven approaches you can take to increase your odds of success. Iavor Bojinov, assistant professor at Harvard Business School and former LinkedIn data scientist, breaks down five critical steps for an AI project to turn into an effective product: selection, development, evaluation, adoption, and management. He’s the author of the HBR article “Keep Your AI Projects on Track.”
Get the Snipd podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share & Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode