“Web 3.0” -- or the “semantic web” -- is the buzz of 2018. Machine intelligence (A.I.) and distributed ledgers (blockchain) are components of this
internet evolution, and higher education professionals are being bombarded with hype from researchers, bloggers, and vendors about how these new technologies can solve all of their problems.
Understanding and leveraging these technologies to serve learners, educators, and employers is the driving mission of the
T3 Innovation Network, a joint project of the U.S. Chamber of Commerce Foundation, the Lumina Foundation and other stakeholders, including tech vendors, U.S. higher education institutions, and employers. And while they hold great promise, there is still work to be done.
“Many higher education professionals are seeing these opportunities and are excited about what they offer, but institutional leadership may not understand the potentials and risks associated with these new technologies,” said Matt Gee, CEO at
BrightHive.
These three emerging technologies can “drive the entire ecosystem forward,” Gee said, if higher education can fully grasp the state of the art and adequately address the challenges that come with them.
What’s the semantic web?
Fundamentally, the semantic web is about moving from cloud-based spreadsheets and shareable documents into more deeply interconnected and openly available data.
“Being linked and being open are so helpful because that’s what human knowledge is ultimately about -- relationships between ideas,” Gee said. “If we want all of human knowledge to be available at our fingertips, it needs to be openly available and linked.”
Instead of links to websites, Web 3.0 offers a deeper and richer level of connection across the internet.
To understand the implications for colleges and universities, Gee said, imagine googling nursing programs in Denver.
“Instead of just having all available programs come up, you could actually search for particular credentials or specializations, the teaching methods at those institutions, and get personalized recommendations based on what you already know -- your current resume, current transcripts, etc. -- and discover which program will add credentials and competencies you don’t have yet,” Gee explained. “Right now, you can’t do that because not all of the data is openly available and linked on the web. But we can get to it in the next 10 years -- in part because of the great work being done by people in the T3 Network.”
In addition to the “semantic web,” Web 3.0 also promises major advances in artificial intelligence and distributed ledgers.
Artificial intelligence and equity
Gee’s background as a data scientist means he’s familiar with machine learning and the all-powerful algorithm. AI may soon make simple tasks, such as identifying skills and competencies in a course catalogue, and more complex tasks, such as recommending career paths, more efficient and effective, but AI is only as smart as the people who make the rules.
“These applications of AI rely heavily on the first technology [semantic web], because in order to help algorithms get better, we need open, linked data in a machine-readable format on the web,” Gee said.
However, machine learning relies on replicating previous decisions, so any bias in past data will only be reinforced by algorithms unless those biases are corrected.
“Although we have visions of machines that are smarter than us, they’re actually pretty dumb -- they just take a look at things that have happened in the past and create a set of decision rules to replicate that exactly,” Gee said. “For example, if you feed the algorithm average descriptions of doctors and nurses from across the web, you end up with unfortunate implicit biases that associate doctors with male identity and nurses with female. That’s in the underlying math and it can mean that if we aren’t careful with the AI, an application will recommend that men go on to become doctors and women go on to become nurses.”
This cautionary tale demonstrates how important it is that practitioners are aware of implicit biases in their data and are asking their partners and vendors the hard questions about how they are dealing with access and equity in their data analysis and correcting those biases in their algorithms.
“You have to do it deliberately and carefully,” Gee said. “But, done right, using AI can actually dramatically improve outcomes and equity for those who have historically been disconnected because they can get improved recommendations and tailored advice even without having access to advising and well-moneyed institutions.
“Equity and bias are some of the most important and most challenging issues to address in the world of machine learning and AI,” he said.
Architectures of trust: Privacy and identity control
Another key issue institutions will confront with these technologies is who has access and control over an individual’s information.
“Our society needs to ensure we have control and sovereignty over our own information, and there are great new policies like GDPR pushing in that direction,” Gee said. “How do we go about it so that every individual, not just those most well off and most digitally savvy, has control over their own data? There are a variety of new architectures of trust.”
Distributed ledgers (like
blockchain) are tools that can help ensure individual control over data in a way that also helps to personalize the user’s experience. For example, a student’s journey from high school to a two- and/or four-year institution into graduate school or work is an individual, not aggregate, journey -- and the data and output should be as customized as that experience.
"One really exciting potential use of blockchain is the ability for an individual to have greater control over their own data and have that data work for them in beneficial ways,” Gee said. “What these groups of technologies generally do is allow for a variety of individuals and institutions to share a set of verifiable, secure, and encrypted information across a network of individuals and institutions.”
The ability for information to be both in sync and secure across a variety of different places means that a network of institutions (distributed) have individual-level records (ledger) regarding the academic achievement of a particular individual. The record may be partially housed at multiple institutions -- high school, community college, four-year, and so on -- but for the individual, it acts as a single record over which the user has control.
Become informed and contribute to the conversation on your campus
In a Monday morning session at the 2018 AACRAO Technology & Transfer Conference, Gee will lead the important
Innovation Hub sequence session “Demystifying Today’s Tech Buzzwords: A Hitchhiker's Guide to AI, Blockchain, and Semantic Web Technologies in Education.”
“We’ll discuss concrete use cases, talk about the challenges and limitations of these technologies as they stand and discuss folks’ questions, concerns, and successes,” Gee said.
Attendees can expect to take back to their campuses clear information that will support more informed discussion about the roles these three technologies can play at their institutions in the coming year.
Learn more about the 2018 AACRAO Technology & Transfer Conference, July 8-10, in Minneapolis.