5 Artificial Intelligence Business Lessons From The Field
The big data and analytics market continues to morph as Artificial Intelligence (AI) fields, such as machine learning and deep learning, provide new ways to generate business insights. O’Reilly’s Strata Data conferences showcase how AI technologies are changing what companies can do with data across a wide range of industries. Here are four themes from Strata Data that universally apply across various industries and company sizes.
1. Both humans and machines are needed to deliver the best result.
Pinterest’s SVP of Engineering, Li Fan, shared the challenges and opportunities of creating a visual discovery engine. Fan described the challenges with naming an image while also understanding what’s in it and the style behind the image. Correctly defining image attributes is crucial for delivering a successful user experience. For example, a living room has multiple items in it. The company uses computer vision technology to break down the image, understand the objects within the picture and recommend similar things for you to consider.
Since there are 100 billion pins in the database, Pinterest can’t rank every pin for every user in real time. To accomplish this herculean feat, Pinterest uses a graph-based recommendation engine that filters the candidate recommendations for every user. It uses machine models to predict the engagement level of a Pinterest user to a pin and the relevance of pin for a user. It sounds like a simple classification problem, but the system not only has to detect an item, such as a chair, within millions of images. Additionally, the AI system has to understand the style of the object, requiring Pinterest to create feature vectors to help recommend an image based on a user’s style. Where does the human element come in? Data specialists clean, validate and label the data.
Every night that data is fed into a model and retrained to improve the result. It’s a partnership between Pinterest employees and computer algorithms. Other sessions discussed how people are working with machines to classify data but also to review the results. Takeaway: There’s a role for human plus machine collaboration. In fact, Paul Daugherty, Accenture’s chief technology & innovation officer, has written a book on this called Human + Machine: Reimagining Work in the Age of AI.
2. It’s not defining the algorithms that stall various AI efforts.
Both IBM’s Dinesh Nirmal and O’Reilly’s Ben Lorica said preparing data for mathematical models was the primary bottleneck for AI. Nirmal’s keynote focused on operationalizing AI. Nirmal described how real world machine learning reveals assumptions embedded in business processes and in the models themselves that cause expensive and time-consuming misunderstandings.
Data hygiene has been a critical failure point that has thwarted analytics efforts since the dawn of time. However, it’s an even bigger issue as companies look to incorporate lots of data from various internal and third-party databases. IBM talked about the need for preparing data but also having a structure for AI model management. In a meeting with Ben Lorica, he noted there’s a role within the AI/data science discipline called data engineer that assists in preparing data for the data scientists to use in the algorithm training process. Even in 2018, we’re still trying to eliminate the garbage in yield garbage out problem.
3. AI will impact jobs in every industry, across all roles.
Michael Chui from McKinsey’s Global Institute for Growth described a world where AI changes every category of employment. He noted that while AI might not replace a person’s entire job, it will eliminate repetitive tasks within all roles. Minimizing repetitive tasks will enable a company’s employees to focus on higher level functions.
I agree with this sentiment. However, this outcome will only be positive if individuals, and the companies that employ them, focus on new skills development. It’s unlikely that employees will be self-motivated to learn new skills during nights and weekends. It will require a dedicated effort from both Human Resources and senior leadership to integrate technology education into an employee’s weekly workload. Retraining the workforce should be a top priority within organizations.
4. Don’t fall into the one tool to rule them all trap.
If you buy a hammer, you want everything to be a nail. Tobias Ternstrom Product Manager, Microsoft described a constant challenge that every technology buyer faces, regardless of industry. Companies want to buy a single tool for the job. Ternstrom described the tyranny of trying to use one software or algorithm for all of your data problems and reminded the group that the best results come from investigating all options and matching the right tools to the right job.
5. Transfer learning can kickstart machine learning efforts in organizations.
In case you haven’t heard of it, Transfer Learning is here, and it’s fantastic. Companies can shortcut the process of developing algorithms by using a model that was trained for a specific task as the starting point for developing a new model for a different job. For example, Ayin Vala talked about how the Foundation For Precision Medicine used ImageNet’s image recognition algorithms as a starting point for designing an algorithm that discovers Alzehiemer’s traits in MRI imagery.
Transfer learning is used to speed up training and improve the performance of your deep learning model. Where do you find these models? One place is in the cloud. Vendors such as IBM, Google and Microsoft are offering pay-as-you-use, pre-trained models as the starting point for computer vision and natural language processing tasks.One thing is clear, while technologists have been on an AI development journey for years, we’ve made tremendous progress in the past five years.