Just years from now, AI will likely penetrate every industry imaginable. The IDC forecasts that by 2025, 75% of businesses will invest in employee upskilling because of skill gaps from AI adoption. It’s an apprehensive thought for our future workforce, of which a sizeable number would inevitably be subjected to automation-led displacement. On the other hand, those within—and entering—the AI industry are on the winning side. Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) is charting the right path as the world’s first graduate-level, research-based AI university. TAIGER had the honour to speak to its first cohort of students in a closed-door Employer-Led Webinar held recently, and share our industry perspective and tips on navigating the landscape as a student and future changemaker.
We wrapped up an employer-Led webinar at MBZUAI
Our Director of Growth, Kevin Quah, attended the event as a panelist and leading AI expert representing TAIGER and the Asia Pacific region. He was joined by three other panelists, Athira Prajith Nair, Bushra Mahmood and Abbas Raza Ali hailing from renowned IT companies in the Emirates. The speakers covered extensive ground together with moderators and hosts from the Student Careers and Alumni team at MBZUAI.
Attendees consisted of MBZUAI’s first batch of graduate-level students specializing in Machine Learning and Computer Vision. An insightful agenda was set, particularly revolving around noteworthy industry applications and an analysis on the future of AI and work. The panelists also included a personal sharing on their journeys into their AI careers and practical tips and suggestions for MBZUAI students. In the hour-long discussion, there were three main talking points that stood out:
AI’s industry omnipresence can no longer be underestimated
As the panelists came together to share their own AI projects, past and present, it was salient that the breadth of application of AI was far-reaching and burgeoning. TAIGER, for example, has worked closely with government agencies from aviation to the arts to deploy our semantic search engine. Omnitive Search was interestingly implemented in our local aviation authority to compare the clauses within civil aviation regulations from different countries through the extraction of the semantic meanings of each clause. The other panelists also quoted niche case studies such as traffic signal management and port operations management; these solutions solve congestion challenges through a range of technologies including predictive analytics. In light of the coronavirus, another use case is applying AI In healthcare to derive explainable AI-driven prognostics of Covid-19; through analyzing chest X-rays or CT scans, doctors are able to diagnose the patients, quantify the severity of the diagnoses, and further extrapolate their historical data to suggest treatment methods for other patient groups.
The takeaway? AI’s broad applicability is opening up many doors for students. The maturing technology is making it possible to realize seemingly absurd AI applications in the near future. In addressing a question on how to find inspiration for research topics, Kevin points out that it is perhaps a matter of simply delving into new industries that one wouldn’t expect to currently use AI. He uses agriculture as an example, where computer vision is being explored in determining soil fertility. One of our panelists, Bushra, also recalls that “Once upon a time it was a sci-fi fantasy to have autonomous vehicles being seen on the road. Now, it’s a reality.” We have progressed from roads, now to agriculture, and will continue to leap beyond—students should dare to imagine such kinds of dreams, especially today, as the technology blooms.
AI starts and ends with people
Coming from the experts, AI always revolves around people—be it our workforce, customers or engineers. Gone is the notion that AI would eventually unfold uncontrollably and take over humans. AI is 100% man made, and much ownership must be assumed by our engineers and data scientists. To progress the field, researchers and students must be discerning in where to invest their effort. Budding researchers should dive into emerging areas—like meta-learning and explainable AI—on top of well-established or popular concepts like computer vision. Ali specifically recommends tackling both experimental and maturing AI in the research field—even hybridizing them if possible, which culminated in his work in both deep-meta learning, and meta-reinforcement learning.
At the same time, where AI ends is where it brings value. The importance of value creation for the end user was repeatedly emphasized by the panelists. Intelligent AI solutions should conclude with the mission to empower workforces or bolster customers in their customer journey. Ali uses an example of goal-oriented conversational agents that drives home this point: where chatbots are often assessed according to their response optimization capabilities, the end objective of the initial query could end up neglected. “If your bot is answering nine questions (out of 10) but not solving the problem, it’s a useless bot. The idea is to optimize the goal—the user problem—rather than answering the question,” he says.
As a closing tip for students, Kevin further introduces the issue of ethical AI, advising students that developers must always consider how the solution will impact people—and whether it is for the good of everyone. He quotes Google’s infamous unofficial motto, ‘Don’t be evil’, as an apt reminder. The rapid advancement of AI will categorically be accompanied by vast ethical ramifications. As AI eventually becomes a core component of our everyday lives, developers in the field must adhere even more strongly to the key pillars of AI ethics.
On top of acknowledging AI as the next big thing, students must stay practical and build key skills
Riding the AI wave as a changemaker is thoroughly exciting, but the panelists caution students to not overlook the important fundamentals in building a strong AI career. Foundational topics like linear algebra, calculus, probability and statistics are crucial to adeptly harness Machine Learning and computer science knowledge. Exposure to applied AI beyond academia would also stand to benefit students. Where classrooms would dive deep into what AI can do, AI firms and labs would reveal the very limitations of the technology, especially when faced with an unfiltered list of workplace or customer pain-points. Academic concepts might deal with simplified and clean data sets like concise excel sheets, for the purpose of exemplifying the AI concepts in question. Yet, realistic industry applications deal with highly unstructured and unprocessed information that warrants extensive effort in data wrangling—and this is only possible after a tedious data collection process. Adding on to these challenges are client expectations of robust models with lofty accuracy benchmarks. The main takeaway is to not belittle the intricacies in building a trustable knowledge stack; go back to basics, but also have the initiative to join research labs or gain industry experiences with AI firms.
A big thank you to the Student Careers and Alumni team at MBZUAI for hosting the employer-Led Webinar, and the various panelists for a conducive discussion. Returning back to the point we opened with, embarking on a journey into AI is a choice to be on the winning side of history. We are excited to invite the first global batch of students of MBZUAI to join the growing AI ecosystem and are looking forward to future opportunities to collaborate together in this game changing industry.