Why future engineers will need more than technical skills

Professor Yusuf Leblebici
President of Sabanci University
Turkiye/ Fellow, IEEE

As AI automates more aspects of design and research, Professor Yusuf Leblebici argues that engineering education must focus less on narrow skills and more on adaptability, collaboration, and interdisciplinary thinking.

How do you see AI tools reshaping the way subjects such as circuits, systems, semiconductor devices and microelectronic design are taught at universities?

These subjects traditionally rely on a solid foundation of physics, quantum mechanics, and materials science – along with linear algebra and differential equations – all of which are taught through relatively “traditional” curricula at universities. Students must understand how a transistor works, based on carrier transport mechanisms under an electric field, or how the network equations describing current and voltage variations are solved, based on fundamental linear algebra. The way these foundations are taught is not likely to change significantly, even with the widespread use of AI tools in education – simply because these foundations are rooted in natural science disciplines such as physics and mathematics.

However, the methodologies used for dealing with design complexity, such as creating complex digital and analog systems consisting of billions of devices, will definitely undergo a fundamental change thanks to AI. Large-scale integrated circuit and system designers have already relied on “heuristic” algorithms for more than 30 years to help them achieve near-optimal design solutions, but the advent of AI tools may open up completely new avenues for complex system optimization. In the domain of integrated microelectronic systems, the so-called “design space” can be enormous even for the most experienced designers to master and to explore fully, and many designers would rather resort to tried-and-tested methods that are known to produce acceptable results. With the help of AI tools, integrated systems will likely benefit from innovative solutions that exploit previously unexplored design approaches that improve performance.

How should educators balance teaching foundational theory versus leaning more on automated tools?

Teaching foundational theory can be assisted to some extent with AI-generated examples and visual aids, but educators may still rely on conventional classroom lectures to convey the key concepts related to semiconductor physics, and circuit theory. AI tools will be much more influential in design automation and design space exploration, as well as exploring published research that may have remained relatively obscure given the enormous volume of research published every year. Especially in senior-level and graduate-level courses, asking students to perform comprehensive literature surveys is always a challenge, but AI tools offer an efficient way to sort through the enormous volume of peer-reviewed articles and pinpoint relevant results. Also, certain design “insights” and “best practices” that can only be developed through hands-on design experience can be accelerated through AI-assisted tools.

What role do you believe AI should play in accelerating academic research (particularly in fields such as microelectronics) and where should we draw boundaries?

Successful research in microelectronics requires a combination of (1) deep fundamental knowledge (2) ingenuity (3) hands-on execution (i.e., actually designing and producing working prototype circuits that prove the novel aspects) and (4) meticulous testing and comparison with benchmarks to show the results.

The cycle to produce new and publishable research in microelectronics is particularly tedious because of the last two elements: designing and fabricating test chips, and testing. Publishing results based on concepts and simulation results alone is possible, but usually has limited impact without “silicon proof.” AI tools can certainly accelerate the concept development phase with very comprehensive design-space exploration and simulations, but the real impact will be seen when AI-assisted design tools speed up the tedious process of actually designing test circuits, as well as performing and comparing test results.

We are rapidly approaching this critical point. This could have a democratizing effect on microelectronics research, when the value of the work is judged by the novelty of the fundamental solution rather than the speed and efficiency of producing a working prototype.

It is worth remembering that the field of digital IC design experienced a similar transformation 30 to 35 years ago, with the advent of automated logic synthesis tools and automated P&R (placement and routing) tools. Before the introduction of such design automation tools, a new design had to be created by full-custom layout, and the intrinsic value of research results also (partially) relied on the quality of execution such as area utilization and wire-length reduction at the individual gate level. With widespread availability of design automation tools that can produce high-quality physical design (i.e., layout), the focus has shifted to block-level and system-level innovations because now, almost everybody can produce a sufficiently decent digital block without much difficulty.

How can universities collaborate with companies like Huawei while maintaining academic independence and long-term research vision?

Especially in areas such as microelectronics, industry-sponsored research plays a very important role because there is a very tight link between the intrinsic value of research and its industrial relevance. This emphasis on practical relevance is quite pronounced in microelectronics. The priorities of academic research must therefore be informed by the priorities of industry, and the most effective way of achieving this link is for the universities to engage with industry partners on a long-term basis. Focusing on short-term projects would not produce the same engagement, because the emphasis would be on immediate results, with relatively limited room for groundbreaking innovation. Long-term engagement depends on mutual trust and a framework that guarantees academic independence for universities while also delivering tangible benefits to industry partners.

What, in turn, is the role and responsibility of the collaborating companies in those partnerships?

In order to extract the maximum benefit from engagements with academic partners, the collaborating companies have to be aware that universities operate on a different time-scale than industry. Typical university projects have a timeline of three to five years, because this is the time frame necessary to engage a doctoral student, for example, on a novel subject. This may seem very long from an industry perspective, because in the same time period, a typical company may go through multiple product cycles. But there are also subjects that require systematic, long-term exploration in order to bear fruit, and these are usually best addressed by a well-designed university-industry research framework. In the end, the company will be able to extract valuable benefits from such engagements that are simply not available through short-term projects.

What new competencies should engineering students develop to remain relevant in an AI-driven world, beyond traditional technical expertise?

Engineering education is already changing quite significantly, shifting its focus from specific skills such as traditional technical expertise or using certain tools and methodologies, towards a more interdisciplinary approach that teaches students to develop abilities across domains. The emphasis is no longer on “skills” (because certain skills that seem valuable today may become irrelevant tomorrow), but increasingly on “capability building,” i.e., gaining the ability to adapt, being exposed to new domains and disciplines, and being able to bridge across different disciplines. This also requires stronger social skills, because the students (and later, the future engineers) must be able to speak with practitioners of very diverse fields in order to create common solutions.

At Sabancı University, we emphasize an interdisciplinary approach to education. The university does not have any traditional “departments” but only “diploma programs” that interact with each other, and we encourage our students to take courses outside of their comfort zone to expand their horizons. While the great majority of our students are engaged in programs related to engineering and natural sciences, many also take courses (or have minors) in disciplines such as economics, psychology, international relations, or business administration. This type of exposure allows them to become more familiar with other disciplines, to grow intellectually, and to gain more self-confidence to think and to act in an interdisciplinary manner.