Hot Topics of Generative AI are the Focus of 2023 Data Science Summit
Written by: Evan Koblentz
Artificial intelligence experts from ADP, Amazon and Maersk highlighted this year’s NJIT Data Science Summit, hosted by the university’s Institute for Data Science.
The summit began in 2022 with Google and IBM presenting about hardware and ethics. This year’s focus was on generative AI, large language models and how they’re used in the business world.
“Data science is a new discipline that enables data-driven decisions across real-world problems in areas such as healthcare, security, retail and advertising, human resources, urban sustainability and entertainment,” said NJIT’s David Bader, distinguished professor and director of the Institute for Data Science. “We were delighted to welcome the community to our NJIT @JerseyCity campus where we teach in our data science masters and certificate programs.”
Amazon’s Sherry Marcus, director of applied science for generative AI, told the in-person and online attendees that generative AI has been happening for more than a decade. What is new, she explained, is access to open-source datasets, inexpensive GPUs (a type of computer processor ideal for crunching big data), and the 2017 public description of transformer software, which is a deep-learning architecture relying on parallel processing.
“It’s the combination of these three things that as far as industry work is concerned, allow generative AI to take off,” Marcus stated.
Everyone knows about large language models like ChatGPT, so Marcus instead discussed four other kinds of AI applications that she believes will change how businesses and consumers operate. They are contact centers, code generation, image generation and machine diagnostics. Meanwhile, she said prompt engineering is the easiest way to move a business into the AI age, while building one’s own language model would be the hardest way.
“So right now, the industry is trying to reckon with all of this,” Marcus said. Amazon engineers and business leaders spend a lot of time thinking about AI explainability, fairness and robustness, she added. “As we move on with regulations, responsible AI is going to become more and more embedded within development, within training and within the whole lifecycle of applications. And that is something that we’re just very much at the beginning of.”
Erez Agmoni, global head of innovation for shipping giant Maersk, and Xiaojing Wang, distinguished engineer at ADP, both presented about their own companies’ uses of AI for practical matters. Agmoni focused on AI to help understand the data behind logistics, while Wang talked about user experiences in human resources.
NJIT professors Mengnan Du, Usman Roshan and Mengjia Xu, all from Ying Wu College of Computing, presented research in AI reliability, medical diagnosis and human brain analysis, respectively.
One of this year’s attendees was Jessica Horton, a spring 2023 NJIT @JerseyCity graduate from the online M.S. in data science program. She worked as a software engineer for actuarial and financial modeling and forecasting, before pivoting her career to data science.
“I’m really excited about advances in AI with image recognition in medical imaging, for trying to recognize tumors and things,” Horton said, adding that Maersk’s Agmoni was her favorite speaker of the day. “It’s really interesting seeing the difference between research for pure research sake, and then how it’s applied in the real world with a company like Maersk and how they spent months just collecting data, not even looking for a specific answer, to see where the data could lead them.”
Horton said she highly recommends the NJIT data science program because of the knowledge and connections gained from faculty. She cited Associate Professor Yiannis Koutis in particular. “The way he explained everything, it was just so easy to follow … with the examples that he gave, being able to dive in and do some coding.”