�������ϲʹ���ֱ��

�������ϲʹ���ֱ�� computer science alumnus, Gang Chen, M.S.’98, is making strides in some of the fastest moving industries of artificial intelligence (AI) and machine learning (ML). After gaining over 20 years of experience as a software design engineer and later a principal engineering manager at Microsoft, Chen now works as a senior manager in Amazon’s Artificial General Intelligence group managing several engineering teams that work in key areas of generative large language model (LLM) initiatives.

“The actual technology I learned in college has progressed, but the methods of learning and researching will stay with me lifelong. That’s the beauty of college education, it’s not just an algorithm you memorized, or a math problem you solved, but rather the methodology and learning habit you formed along the way. In machine learning, we have the same challenge. We don’t want to teach every tiny detail to LLM, and it would be bad for LLM to overfit and memorize things. We find ways to instruct and teach LLM the reasoning so that it can solve a class of similar problem instead of a particular one.”

What are the top trends in AI and how will they impact organizations?

Chen: AI will impact people’s lives in many ways, and we haven’t figured out everything yet. I personally found a few areas particularly interesting:

• Enterprises using LLMs:Every large company has a huge amount of corporate data they would like to mine and utilize to better serve their customers, and most enterprises are reluctant to provide that to LLMs, due to concerns of confidentiality. There are many services and applications that can be built around this, for example, a call center and customer service would benefit from extensive knowledge of the product catalog and details.
• Multimodality: We have seen handwriting, touch screen and voice input becoming a trend in consumer electronics. We even saw applications telling you what the plant/product is in the picture you took. Humans can interact through different ways such as speaking, writing, drawing and acting. Machines should be able to do so as well. We will see a lot of AI applications interacting with users through multi-modal inputs.
• Device computing:The edge device is getting more and more powerful these days. Apple silicon now has multiple central processing unit and graphics processing unit cores embedded. Instead of sending every input to the cloud, personal computing devices such as your Mac or iPhone can host LLM models and handle requests locally, saving costs, improving latency, and protecting privacy.

What are you most excited about in your field?

Chen: The field of natural language understanding has been turned upside down since ChatGPT was released. Big companies such as Google and Meta are racing to release their own LLMs while Microsoft established a multiyear, multibillion dollar partnership with OpenAI. I am super excited about AI’s potential and believe it will help drive the industry to a new high much like PC dominance, cloud computing and mobile device. Not only LLM, but also sensory technology, robotics, and an entirely new mindset of living day-to-day with AI assistance. 

What do you see as the biggest challenge today?

Chen: I think Responsible AI is today’s biggest challenge. Responsible AI is about fairness, privacy and ethics. It is extremely important to consumers and users but does not necessarily align with a for-profit company’s short-term goals. Google, Microsoft or Meta can spend billions to secure graphic processing units and train large models; however, no university could do the same. This greatly limits academia, which is usually less financially focused and more socially aware, in  their ability to stay on the front of state-of-the-art technology and help steer the direction

What is a tip you would share with students about success in school and launching their careers as engineers?

Chen: Learning is a lifelong experience. You need to have intensive curiosity and learn continuously, especially if you want to continue your career in any STEM majors, as these areas move fast.

Also, don’t be satisfied with your learning on the job through the project assigned to you. These sometime stay repetitive. Branch out, learn from open source, learn from industry, learn from online resources and build a network with friends. Keep finding new challenges and don’t get comfy and settle. Looking back on my own career, I probably could have moved a bit more often and settled less.

For more �������ϲʹ���ֱ�� news, arts and events, visit �������ϲʹ���ֱ�� News online.