There is a Jataka tale about a king who had a wise parrot that had travelled widely and could name the failings of every minister in the court. The king consulted it on everything. Until the parrot, swollen with importance, suggested it might sit upon the throne while the king attended to lesser matters. The king, who knew the difference between a clever voice and a crowned head, replied: “You may speak from the perch; you may not speak from the throne. “
I read this story alongside an unusually careful institutional note now circulating among Higher Education Institutes (HEIs) in India. It proposes that every Board of Studies — the committee of professors and industry experts that decides what a degree shall actually contain in terms of courses and content — begin making structured use of an AI tool. Not as a member. Not as a voter. But as what the note describes as an “AI Special Invitee”, a guest with a perch and emphatically no throne. The proposal is quietly sensible but also easier to write than to honour.
The Board of Studies is one of those bodies the public never sees and the academic never forgets. It is where a syllabus is benchmarked against comparable programmes, where outdated content is pruned, and where someone finally asks whether the way students are examined bears any relation to what they were meant to learn. Get it wrong, and a cohort of graduates will carry the error into the workforce and the country will have to live with the consequences for decades.
The note would let AI assist here by comparing a draft curriculum against national and global programmes, flagging gaps and redundancies, checking whether outcomes align with assessment, and summarising scholarship that has moved faster than the syllabus. Useful but tedious work. The note is scrupulous about the perch: the tool does not vote, is not counted toward quorum, and holds no decision-making authority. Every academic judgment stays with the duly constituted members. The machine may speak; it may not decide.
This is the king and the parrot, rendered into administrative prose. In principle, it is right. The trouble is that such principles tend to be strongest on paper and weakest at four in the afternoon, in the third hour of a meeting, when a tired committee is handed a fluent, confident-looking summary.
A tool cannot walk in and present a slide. So, each Board must designate a faculty member as the “Operator”, who runs the queries, prepares the analysis beforehand, shares the screen transparently, and represents the output faithfully. Readers will recognise this figure. He is the mantri, the minister who gathers counsel and carries it to the one who must choose. Tenali Rama beside Krishnadevaraya, Birbal beside Akbar: their genius was not that they ruled, but that they brought the right information to the person who did.
The note insists that accountability rests with the Operator and the Board and not with the tool. This seals an escape hatch we have grown far too fond of: the loan refused by “the system, the application rejected by “the software”. Here, a human prepared the analysis, a human presented it, humans decided. No one may hide behind the parrot.
But notice what the Operator is being asked to be: both the enthusiast who runs the tool and the sceptic who must catch its errors. Those are different temperaments and we are asking one tired colleague, often the convener already burdened with minuting the meeting, to hold both. The mantri who is also taking dictation is not in a strong position to interrupt the king.
The framework leans, finally, on a single sentence repeated in various forms: AI outputs must be verified before they are relied upon. Everything safe about the scheme hangs on those five words. Verification is precisely the step that institutional life in HEIs is worst at sustaining.
Consider what honest verification of a benchmarking summary would require: pulling the actual syllabi of the programmes the tool claims to have compared, confirming that the regulatory norms it cites read as quoted, checking that a confidently named “recent development” in the discipline is real and not a plausible invention. That is hours of work, which the tool was brought in to save. The danger is not that AI will seize the throne. It is subtler and more human: that its smooth output will be waved through because checking it is dull, slow, and socially awkward, and because it usually looks right. A wrong figure delivered hesitantly gets questioned. Committees are not seduced by accuracy; they are seduced by fluency.
The parrot, remember, was dangerous not when it was wrong but when it was charming. An AI summary is charming by design — fluent, organised, shorn of the hedging that betrays a human who is unsure. “Verify everything” asks a committee to be most suspicious of precisely the input that is easiest to accept and to spend its scarcest resource — attention — on the document that least seems to need it. That is not how committees, or people, actually behave.
There is a further trap the note does not name. Once a tool is approved, official, and screen-shared, its output acquires an authority the same words would never carry from a junior colleague. A draft analysis that arrives stamped with the institution’s blessing is half-accepted before anyone reads it. The framework intends the AI to have no greater weight than any other advisory submission; in the room, the opposite pressure quietly operates. The mantri who speaks in the king’s own voice is hard to contradict, even when he is only repeating what the parrot told him on the way in.
None of this is an argument against the experiment. It is an argument for being honest about where it can fail so that the design can guard the right door. If I were advising a pilot Board, I would insist on three safeguards. First, separate the roles: the colleague who runs the tool must not be the one certifying its output. Second, no AI-generated fact should enter the minutes without a checked source. If it cannot be verified, it must be struck. Third, log not only where the tool helped but where it erred. A pilot that reports only time saved and never an error caught has not verified anything; it has merely trusted quietly.
Why should this matter to a reader who will never sit on a Board of Studies? That is because the question being negotiated in this modest note is one of the decade’s central ones: how does an institution absorb a genuinely powerful instrument without quietly outsourcing its judgment to it?
Two easy postures are on offer and both mistake the parrot for the king. One treats AI as an oracle, accepting its answers with a reverence once reserved for scripture. The other treats it as a contaminant, to be kept out of every serious room. One would crown the bird; the other would cage it. The harder path and the one the note reaches for and very nearly secures is to let the clever voice speak from the perch while keeping the throne accountably human. That only holds if the humans do the unglamorous work of checking and keep doing it after the novelty fades.
The knowledge gap is not the distance between what the machine knows and what we know; the machine will always recite more facts, faster than any committee on a Tuesday afternoon. The real gap is between information and judgment; between the voice that recites and the authority that chooses and is answerable for the choice. The parrot may have travelled the world, but it is still the king who must decide where the kingdom goes and stand by the decision when the kingdom asks why. India’s universities are about to test whether they can keep that gap honest — not in the founding document, which is admirable, but on the dull afternoons, when no one is watching the perch.
Views expressed are personal
The writer is the Vice-Chancellor of RV University, Karnataka.
Published - July 13, 2026 08:00 am IST