Basudev Biswal

Hydrology. Geomorphology. Environment. Science. History. Philosophy. IIT Bombay. Professor. Views are personal.

[Basudev Biswal @ProfBasudev]

We need to address the elephant in the room: science has theological roots. https://t.co/lflItT3GFB

[Basudev Biswal @ProfBasudev]

@sanjeevsanyal @DivaJain2 You are giving someone the idea.

[Basudev Biswal @ProfBasudev]

"In some sense, we are back to the age of research."

Inevitable. Let history repeat itself.

[Karl Mehta @karlmehta]

Ilya Sutskever says the AI scaling era is running out of road because pre-training is about to hit a finite-data wall.

"At some point though, pre-training will run out of data. The data is very clearly finite."

"Then, okay, what do you do next? Either you do some kind of a souped-up pre-training... or you're doing RL, or maybe something else."

"In some sense, we are back to the age of research."

His timeline is the part most AI investors and founders should hear:

"From 2012 to 2020, it was the age of research. Now from 2020 to 2025, it was the age of scaling... But now the scale is so big."

"Is the belief really that... if you had 100x more everything would be so different?"

"I don't think that's true. So it's back to the age of research again, just with big computers."

[Karl Mehta @karlmehta]

Ilya Sutskever says the AI scaling era is running out of road because pre-training is about to hit a finite-data wall.

"At some point though, pre-training will run out of data. The data is very clearly finite."

"Then, okay, what do you do next? Either you do some kind of a souped-up pre-training... or you're doing RL, or maybe something else."

"In some sense, we are back to the age of research."

His timeline is the part most AI investors and founders should hear:

"From 2012 to 2020, it was the age of research. Now from 2020 to 2025, it was the age of scaling... But now the scale is so big."

"Is the belief really that... if you had 100x more everything would be so different?"

"I don't think that's true. So it's back to the age of research again, just with big computers."

[Basudev Biswal @ProfBasudev]

The Real Threat of AI is Monopolization of AI, not Use of AI for Essay Writing

Imagine you used a computer to do a calculation that was difficult with a pen and a paper, and later you were told the work did not belong to you. Does it sound ridiculous? Well, that is what the warriors against plagiarism, paranoid by AI bots, are telling us. To understand this, let us break down the problem into two parts: i) detection of AI use and ii) ethics of AI use.

A professor of Texas A&M University in 2023 threatened to fail his entire class, accusing them of using AI for writing essays. Fortunately for the students, social media backlash forced the university administration to review the case seriously. Most student essays were later found to be falsely flagged by the AI detection tools.

Why do AI tools raise false flags? Well, AI bots generate sentences and paragraphs by predicting one word after another (they actually predict token by token, but let’s not complicate it). Since the prediction is done probabilistically, pretty much the same way we humans do, texts generated by AI bots almost certainly pass any traditional plagiarism tests. Or, at least, they will do the job as good as humans can do. After all, AI chat bots have been designed to mimic human writing.

For detecting AI generated texts, AI detection tools instead look at the probabilities of word usages and sentence structures. Since there are not many AI bots in the market, it is quite easy to know the probabilities of certain word usages and sentence structures of an AI bot. For example, even without using an AI detection tool, one can quite easily tell if a paragraph has been generated with ChatGPT or not by looking at the number of times the word “delve” and the punctuation “em-dash” have been used.

So you have bad luck if you have a writing style that matches that of an AI bot, and AI detection tools will not show any mercy before branding you a plagiarist. AI detection tools will even flag the texts from the US Constitution as AI-generated, which is an impossibility, by 237 years in magnitude. Fortunately for the students in the Texas professor’s class, they had solid evidence such as Google Doc time stamps to prove their innocence.

Coming to the second part of the problem, a few students in the class admitted to having used AI bots for writing their essays. Disciplinary actions were taken against them as per the university rule, whose outcomes are not in any public domain to my knowledge. I won’t comment on the teacher’s right to punish the students because I am a strong believer in the idea of instructors having the freedom to decide their methods of evaluating student learning. I won’t blame the professor if he had declared his evaluation policy at the beginning of the course.

Nevertheless, the real question is whether the use of AI bots to write an essay should be considered ethical or not. Coming back to the original argument, AI bots are essentially doing calculations for you pretty much the same way a computer would execute the program written by you to calculate the tax that needs to be deducted from your income in a given year. The words in your prompt are mere numbers for the computer with certain probabilities of occurrence.

That AI bots for essay writing sound unethical is largely because it requires so little effort. Not so long ago, people viewed the use of computers as “mechanization of the soul,” an existential crisis. The panic against wires and chips has deep European philosophical roots that held humans as the only rational beings capable of executing complex tasks. The outrage against the use of AI bots to write essays in classrooms is thus not very different from the outrage against calculators in classrooms back in 1986.

I do not want to say that there are no concerns against the use of AI bots for writing essays. However, the concerns are not very different from those against the use of computers for performing calculations. Pretty much like how submission of input data and codes is mandated to address the reproducibility crisis, we can ask students to submit all the prompts and the other input data for addressing the unethical use of AI bots.

Now the interesting question is who is benefiting from the mass panic against AI bots? The answer is actually the AI companies themselves. By flagging an essay as plagiarized, the AI companies are, knowingly or unknowingly, trying to convince us that any style of writing adopted by an AI bot is proprietary material. The AI companies did not invest much creativity to build their bots. All they invested is massive compute hours; the inputs were taken from us only, even without our permission.

Using AI detectors to punish students for writing essays is essentially voluntarily giving monopolistic advantages to AI companies, which otherwise would not be possible for them to obtain.

© 2026 Basudev Biswal. All Rights Reserved

[Basudev Biswal @ProfBasudev]

@svembu Quite true. More data is not necessarily good, and how data is used is more important.

[Basudev Biswal @ProfBasudev]

@PrinceArihan You are exceptionally courageous.

[Basudev Biswal @ProfBasudev]

@theskindoctor13 Protesting is a good business.

[Basudev Biswal @ProfBasudev]

Why is India lagging in innovation?

Why don't our entrepreneurs build frontier AI models?

Who is to be blamed? Scientists? Entrepreneurs?

Actually, a more relevant question is: what is the incentive for model building?

The harsh reality is that most people, including trained researchers, find it hard to differentiate between model development and model implementation, and those who take the path of model development often struggle to get recognization.

I am not saying this is a problem specific to India. Model developers worldwide face difficulty in getting recognition. However, there is an additional difficulty in India: recognization from Indian researchers is conditional -- the contribution should first be recognized in the West.

And the primary reason is that we have willfully handed over the job of judging science to scientists in the West (journal editors).

C. V. Raman recognized the problem of over-relying on foreign publishers. He started journals to address the problem. Unfortunately, scientists in India could not overcome their colonial hangover.

So before asking why Indian scientists do not build frontier models, let us fix the incentive structure.

Starting our own publishing platforms would be a good starting point for the long journey ahead.

(Image generated with Gemini)

[Basudev Biswal @ProfBasudev]

@sanjeevsanyal 😀

[Basudev Biswal @ProfBasudev]

@sanjeevsanyal Getting recognization in India for original work is difficult.

[Basudev Biswal @ProfBasudev]

@tehseenp It seems the ethanol policy has less to do with the import issue and more to do with the excess farm production caused by subsidy schemes. It is essentially a band-aid over the water management disaster caused by poor agricultural policies.

[Basudev Biswal @ProfBasudev]

Does not make a lot of sense to me. Rajastan and West Bengal cannot be put together. Also looks like the flow of water is not taken into account. People living close to rivers will have more water security than those who live far away.

[Joseph T Noony @JoeAgneya]

'Why don't you want IUML handling education ministry in Kerala?'

Well, for starters...

They celebrate Tipu Sultan. The barbarian who pillaged my home district and destroyed the Kerala School of Mathematics. The last indigenous scientific effort in India. Tipu destroyed the legacy and destiny of Kerala. The last Kerala Mathematician Sankara Varman had to flee for his life from Malabar to British Madras. Even Madhava of Sangamagrama's personal Krishna shrine was burned down and the family repaired it only a century later.

You want these guys to control the Education system of Kerala? Do you think they care about the intellectual and philosophical legacy of Kerala? No Malayali with historical literacy and self respect can forgive this cave man with a crown. But then again, Malayalis do not have literacy in things that matter most long term.

Won't be long before you see official Tipu Jayanti celebrations in Kerala. Bright future ahead! Intellectual destiny lost. And now demographic and political destiny also in free fall...

[Basudev Biswal @ProfBasudev]

@MumukshuSavitri Revealing.

[Sanjeev Sanyal @sanjeevsanyal]

Kharavela, who avenged Ashoka’s genocide and ended the Mauryan empire, will now be given due importance in Odisha

[Basudev Biswal @ProfBasudev]

@svembu @garvit_sethii @SanjayTweets9 Yes, the correlation between exam scores and success in real life is very poor. Generations who have grown up being told otherwise will find it hard to accept.

[Priyanka Chaturvedi🇮🇳 @priyankac19]

Imagine spitting at the sun and think the spit has landed. BJP attacking India’s great son, Shri Biju Patnaik is a disgrace and an apology is due

Biju Patnaik was an Airforce pilot with Royal Indian Airforce and with Indian Airforce post Independence. When Pakistani backed tribals invaded Kashmir in Oct 1947 Indian Government organised airlift of India troops from Delhi to defend Kashmir . The airlift started morning on 27th Oct after signing of Instrument of Accession . Biju Patnaik was amongst the first set of pilots who landed their Dakota at Srinagar Airport in middle of fight without any intelligence whether the airfield was in enemy's control or still in forces control . This airlift by brave pilots of IAF helped stop the advancement of enemy and saved Kashmir

Biju Patnaik on request of PM Nehru flew to Jakarta and landed on an improvised airstrip using left-over fuel from abandoned Japanese military dumps, Patnaik took off with Indonesian political leaders such as Sutan Sjahrir and Sukarno, for a secret meeting with Nehru at New Delhi reaching India via Singapore on 24 July 1947.

During 1962 war when Sh. Biju Patnaik was CM of Orissa, PM Nehru asked him to help India setup a China monitoring Air Unit at
Charbatia Air Base near Cuttack . This Aviation Research Centre (ARC), a covert wing of the Intelligence Bureau and help with intelligence gathering against China and still functions under R&AW .
Dr. APJ Abdul Kalam and Biju Patnaik shared a crucial partnership in strengthening India's defense, centered on the establishment of the Wheeler Island (later named Dr Kalam Missile Testing Range) missile testing facility near Balasore In 1993, then-Odisha CM Patnaik allotted the land to DRDO on the condition that Kalam would develop advanced missiles there to safeguard India. Patnaik's decision to hand over the land for free to the DRDO was based on his visionary desire to ensure India was robust against external threats. The island became the crucial site for testing major Indian missiles like Agni ,Akash Prithvi, Brahmos etc

And BJP instead of honouring him with a Bharat Ratna has the audacity to call him a CIA AGENT!

[Priyanka Chaturvedi🇮🇳 @priyankac19]

In the zero hour today raised the issue of judicial over reach in the case of banning Professor Michel Danino and two academics from any role receiving public funds for including a chapter on judicial corruption in NCERT Class 8 textbooks

[Basudev Biswal @ProfBasudev]

@ARanganathan72 @narendramodi @ugc_india Relating crime to identity is a dangerous game. It has never ended well anywhere in the world.