This is the first in a series of posts on my lessons learnt from being in an academic environment that promotes a research mindset, and also as an administrator having tried my hand at nurturing and strengthening such an environment.
*~*~*~*~*
Conventionally, the term "research" was associated with universities and academia, while "industry" was all about production, manufacturing, construction, and business. This is because, research is the process of creating new knowledge, and up until the mid 20th century, knowledge was seen as the exclusive forte of academia and universities.
But even this stereotype started changing by the mid 20th century. For example, many foundational concepts that make up a textbook in Computer Science today-- like automata theory and finite state machines, have had their roots in industrial research labs like Bell labs, Xerox PARC, etc.
Ever since the end of the second world war, industries have steadily acquired more and more knowledge processing capabilities. Today, even the conventional industries invest more in knowledge processing than in the actual mechanics of construction or manufacturing. An EV car manufacturer for example, need not learn about internal combustion engines at all, but needs to learn about software engineering for safety critical systems.
The role of knowledge processing in industries have become so much, that some people even argue that universities are not needed anymore, and that (especially with the onset of AI) universities are broken, and not relevant anymore.
There are also several corporate entities starting their own universities stressing on the "industry focused" nature of their syllabus and research activities.
*~*~*~*~*
In this post, I would like to characterise the social value of what we conventionally know as academic research-- knowledge building activities involving more of books and thinking and debating, and less of programming, GPUs and AI.
Firstly, let me address the misconception that academic research is "theoretical" in nature, while industry research is "applied". I think this framing is incorrect. Several theoretical results have come out of industrial research, and a lot of academic research address practical issues.
Yet another misconception is that academic research is somehow "blue sky" in nature-- in that, these are knowledge pursuits as an end in itself, pursued just for the aesthetic beauty of the underlying mathematics. Maybe so, but I find this argument too often being used to evade accountability. "Just give us money and leave us alone" is what this argument seems to say, which after being an administrator, I find it hard to digest.
This post is hence, about the social value of academic research. If society is to support industry or academia, there needs to be a value proposition that we offer to society. As academics, we have to now ask that if knowledge generation is greatly facilitated by the better levels of wealth and resources available in the industry, does academic research have any social value at all?
Similarly, when we say that AI can replace education altogether, and industry can offer more resources than an academic environment ever could, does conventional universities hold any social value at all?
At least in the current way we are interpreting the role of universities-- as institutions providing skilled personnel to the industry-- our answer may well be that universities will be obsolete soon, and the society will not miss anything.
*~*~*~*~*
I would however, like to argue just the opposite. Knowledge processing may have entered the forte of industry and business in a big way in the 21st century, but there is a characteristic difference between how knowledge is processed in the industry and in the university (and no-- it is not applied versus theory).
Here is my understanding about research in the industry and research in academia:
Industrial research builds social capability, while academic research (needs to) builds social resilience.
Industrial research often starts with engineering or business challenges, builds new technologies to address these challenges and sometimes goes further to build a new science that lays the conceptual foundations for the technology. The end result of this is that some form of social capability is enhanced. As a result of industrial research, we typically end up doing something better, faster, at a larger scale, with lesser time and resources, etc. Sometimes the capability advances are so huge, that it ends up "disrupting" the existing way of doing things and bringing in qualitative changes in the way we did things. For instance, the Internet made information flow so fast and so much more capable, that it did not just create faster telephones and telegraphs, or better radios-- it entirely reinvented the way we process information and organise societies.
This kind of research includes both applied and theoretical elements. Many of these could also lead to "blue sky" endeavours that are pursued just for the beauty inherent in these research questions. But all of these primarily are driven by the zest to enhance capabilities.
Academic research on the other hand, had served and will continue to serve, a very different purpose. Academic research is not (should not be) based on what we produce-- but on what we become as a result of the research journey. Academia is fundamentally about building people-- not building knowledge or technology. The outcome of an academic research journey is (should be) greater wisdom, greater awareness, greater philosophical depth in people's everyday conduct.
This could eventually result in better technologies and greater capability, when these people come together and build industries.
Academia builds new knowledge not as an end in itself-- but as a means to build and empower people. It is the way that the people transform when they build new knowledge, which is the key outcome of academia rather than the knowledge itself. The collective outcome from such an endeavour is social resilience.
Social resilience is the ability of a society to absorb shocks and disruptions, and reinvent itself to bounce back stronger. Every increase in capability eventually leads to a critical juncture that results in a paradigmatic change in the way society operates. In such cases, it is our resilience that keeps us from falling into chaos and emerging stronger from the disruption.
Today, AI is disrupting the workforce like never before. There are very real possibilities of large-scale job losses and threats to individual privacy as well as national sovereignty by AI. But advancements in AI continue nevertheless. Will this disruption lead to widespread chaos and strife in our society, or will we as a society absorb this shock and reinvent ourselves, will be a function of how resilient our education has made us.
Social resilience is possible when the individuals operating in it are resilient themselves, and have the depth and capability to reimagine their lives and to reorganise their activities every so often. This comes from the philosophical depth from which they operate. If their ability to generate new knowledge is contingent upon them having access to huge resources and latest technology, they may be capable of specific things, but not exactly resilient as individuals.
*~*~*~*~*
Let me explain further with an example.
With great advances in AI, there is now big concern about AI Ethics and Responsible AI. But the way this topic is being discussed today, has two broad categories.
The first line of thought aims to directly look at the way AI is built and used today, and build relevant structures around it to make it safer and act ethically. These could involve building guardrails, RLHF, hard (constitutional) constraints, sandboxing, etc.
The other approach is to address the question philosophically, asking what do we mean by ethics itself? Is it doing something that conforms to expected norms (deontology)? Or is ethics about acting for the greater good (utilitarianism)? Or acting in a way that is mindful of consequences (consequentialism)? Or upholding some desirable characteristic (virtue)? This line of inquiry also asks whether acting ethically is necessarily in conflict with acting rationally to maximise self interest? Is ethics subjective or can there be an objective definition of AI ethics? Can one person's ethical AI be an evil nemesis when seen from another person's perspective? Can a machine be truly ethical or does it only reflect the ethical considerations inherent in its training?
The two approaches are characteristically different. The former approach is addressing an immediate question. It is addressing the question of AI ethics primarily with a motivation to avoid getting hit by lawsuits. And any solution that can deflect the blame away from the AI is good enough for the business. The latter however, tries to understand the concept of ethics analytically and asks architectural questions about AI and tries to understand the meaning of a term like "ethics" that was once reserved for human conduct, to be now applied to machines.
I remember some time ago in a research presentation, a member of the audience expressed concerns about the seriously intrusive nature of the technology being presented and its potential to be misused. For this, the speaker had only responded by saying that these experiments have passed IRB review (ethics board) and hence, it is legitimate. What struck me in this response was that the response did not address any of the specific concerns of future misuse of this technology being developed. It just responded to protect the current activity from scrutiny. The IRB only clears the current, proposed activity, and does not usually delve into its future implications. This episode in my opinion, is a clear marker that distinguishes industrial research (that wants a solution or a patch shipped out fast) versus an academic research pursuit (that wants to understand the deeper, underlying implications).
*~*~*~*~*
Worldwide, we are failing to understand this subtlety, and are increasingly bringing the industrial research framing to evaluate academic research.
I have seen several cases across many institutions, where some highly motivated PhD students would be juggling their research, job, and family all at once. But many of them burn out or drop out in between, even after working very hard and after paying significant tuition, simply because their PhD has an unwritten rule that it requires an "A*" paper to get them to graduate!
What I see is not a failed PhD student or even low quality research. What I see is an individual who was resilient and motivated, but whose spirit was broken because the system failed to recognise the resilience and focused only on the output. As I see it, if this person understood the importance of setting aside time and effort for creating knowledge despite their daily grind and demands from work and family, then even a small success in this process helps them to reinforce their resilience, which would contribute constructively to society in important ways.
But by not recognising this, and by framing research quality through a fairly narrow, and mostly meaningless lens, we are systematically tearing apart the only institution in society where resilience is built.
