In January of 2011, IBM Watson made its debut on Jeopardy; since then the Watson technology platform has been applied to education, customer engagement, financial services, IoT, and the medical industry. IBM made bold claims regarding the business and scientific value Watson’s artificial intelligence (AI) would provide.
Since then, IBM third quarter revenue has declined $390 million ($18.76B versus $19.15B in the third quarter of the previous year). Cognitive Solutions, the business segment which includes the highly-touted Watson, was one of IBM’s poorest performing segments.
In an effort to expand into medical data and medical AI, IBM acquired Phytel in 2015. Following the acquisition, IBM was forced to downsize the company segment; some laid-off engineers spoke up about the platform masking real difficulties in turning AI into a profitable business model despite Watson’s powerful AI under the hood. An anonymous engineer likened it to ‘having great shoes but not knowing how to walk’.
MD Anderson, the University of Texas cancer center, was aiming for their moonshot; using IBM Watson for analytics applications focused on eradicating cancer. As of February 2017, the partnership had fallen apart, costing MD Anderson more than $62MM without meeting their goals.
John Mannes of Tech Crunch states that IBM’s early competitive advantage in AI was brand recognition and longstanding relationships with Fortune 500 companies but in implementation IBM struggled to bridge the gap between the technology and client business needs.
“AI isn’t an amorphous black hole that sucks in unstructured data to produce insights. A solid data pipeline and a domain-specific understanding of the AI business problem at hand is table minimum”
Open source “cognitive computing” has made great strides since Watson’s debut on Jeopardy. Claudia Perlich, professor and data scientist with 6 years of experience at the IBM Watson Research Center confirmed that,
Thomas H. Davenport, author of recent book, The AI Advantage: How to Put the Artificial Intelligence Revolution to Work (Management on the Cutting Edge), cuts through the hype of the AI craze. Davenport explains how businesses can “put artificial intelligence to work now, in the real world”.
Rather than going for the moonshot such as eradicating cancer, Davenport recommends looking for the low-hanging fruit to start. The current state of AI and true value is more foundational than it is flashy and will not replace human interference or judgement. Davenport recommends that companies should develop their own expertise and be more hands-on, rather than looking for the flashy black-box solution.
Engineers from Phytel make it clear that smaller, flexible companies are winning bids for AI projects from their customers: “Smaller companies are eating us alive, they’re better, faster, cheaper.”
Most emerging EdTech companies providing universities with predictive analytics and AI have taken a route parallel to IBM by shooting for the moon right from the get-go. This includes university-wide student tracking systems collecting massive amounts of data and real time early alert systems for identifying at-risk students. According to four anonymous employee reviews of higher education predictive analytics providers on Glassdoor,
“They are also all vision, no execution, constantly chasing the next new shiny object” . . .
“Many products were sold half-baked, implementations took upwards of a year and then rolled out without promised features. Leadership ignored huge gaps in delivery and products. They are out of touch with what is really happening on the front lines” . . .
“By the time customers started cancelling contracts and sales slowed due to poor reputation, it was too late to take moderate action, and a hiring freeze followed by layoffs occurred” . . .
“classic overpromise/underdeliver situation” . . .
If this sounds familiar to you it may be time to take a step back and evaluate the true return on investment that predictive analytics and AI solutions for higher-ed deliver. A simple but frequently overlooked place to start, is quantifying the current impact of student interventions before implementing new predictive analytics driven targeted interventions. This includes quantifying the percent of students retained out of those who received an intervention and the resulting tuition revenue relative to intervention expenditures from current university efforts.
Start by laying a foundation of affordable, easy-to-implement, personalized analytics services that quickly provide a solid, measurement of improved recruitment, retention and financial performance. Not flashy promises but measurable, actionable results for your institution.