AI writing patterns - negative followed by positive

One of AI’s favorite ways of phrasing arguments is to first state a negative, like what something is not, and then follow it up with what it is.

For example, I asked ChatGPT to write something about the greatness of Indian culture. Its very first sentence in response was: “Indian culture is great not because it is ancient, but because it has remained remarkably open to evolution while retaining a deep continuity with its past.”

Another, a few sentences later: “India has never been a single cultural expression. It is a vast mosaic of languages, cuisines, philosophies, artistic traditions, customs, and beliefs.”

And then another: “But the greatest strength of Indian culture may lie not in what it achieved in the past, but in its capacity to absorb, adapt, and renew itself.”

And another: “The greatness of Indian culture, therefore, is not a claim of superiority over other cultures. It is the recognition of a remarkable civilizational resilience: the ability to preserve continuity without demanding uniformity, to encourage questioning without abandoning tradition, and to absorb diversity without losing a sense of identity.”

In my observation this is not the usual way we write. This style is visible in most LinkedIn posts which can be marked “AI slop” right away.

But as humans are reading more and more AI generated content, we are beginning to write like AI as well, and the style is creeping into human-written content, which feels like a degradation of quality if you are particular about linguistic standards.
What are some common free AI style phrasings that put you off?

Heard about the Hugging Face incident?

Any enterprise in the process of transforming their IT by bringing in AI agents must take note of the Hugging Face incident. Driven by goals, and without a handle on the strategic paths these agents take, the agents are capable of behaving in interesting, yet scary ways. Designed to deliver outcomes and moulded by peer dynamics legitimizing jumping the guardrails, AI agents show us how easily they can go out of control.

Must listen, and follow the story.

Does a leader's age matter?

~35 to ~55 is a weird age-range. You just can't tell the age with much accuracy by looking at the person. You may still find out the true age, but what you see is what generally matters, especially at sub-conscious levels, in how you treat and respond to the person. Fitter people look younger, but can't generalize. Being fit seems to matter a lot in generating the right impressions. Height can distort perceptions, despite fitness levels. Gray hair gives an impression of wisdom, but such impressions are often misleading too. Command at language, especially English, can be mesmerizing. Energy levels may not correlate with age and can generate a "wow" depending on how old you think the person is. It seems to me, though, that for leadership positions, if one is in the 30s, it's beneficial if he/she looks like and talks like he/she is in 40s, yet is fit and energetic - like Zohran Mandani. For someone in 50s, it would again be better to look like he/she is in late 40s - as that would reflect energy, wisdom and longer runway all at the same time, albeit the real age - like Shashi Tharoor was for a long time. Obama, interestingly, looked 40s, was in 40s, and behaved 40s throughout his tenure as president. True age, probably, is indeed just a number. But for leaders, there seems to be an optimal age as perceived, which may be different from the true age, and which is a function of a lot of things.

These are just my thoughts, of course, from my 'generative' intelligence on a Tuesday evening, as I struggle to make my daughter study for her test tomorrow. Not sure how I appear to her. Will know some day... or may be not🤪.

Share your thoughts.

Research - a free journey of the free mind

Research is a mindset. It’s an eternally curious state of mind that keeps walking into dark alleys, searching for something - an explanation, a deeper understanding, an inspiration, or plain nothingness that is still beautiful. It is a personal journey, through which humanity has benefited immensely.

Research that has to operate with constraints of methodology and publishability struggles to attain the depths of exploration. Having said that, methodological rigour is meant to establish robustness of findings, which is, without doubt, the most important criteria for acceptability of a research outcome. However, scholars trained to “do” research in text-book ways often fail to appreciate alternative perspectives and approaches on theory, method and robustness. This is especially troublesome in social sciences where experimental evidence is rare and theory building is supposed to be incremental and contributory.

The existing paradigms of funding and incentives around research work have created a culture of risk-aversion. Too much structure and boundedness is not healthy for scientific exploration. Research requires passion that can’t thrive under pressure to generate outcomes.

Agentic AI based Fed

What if the FOMC or the FRB was a bunch of AI agents?

It looks like a great candidate for AI. The goals and objectives of the federal reserve system are clearly laid out. The targets for monetary policy are also well established and agreed. The macroeconomic, fiscal and banking data which the fed analyses is, well, and after all, data. And AI can consume it real-time. All past research has been consumed by AI, with excellent remember and recall. Some of those near-superintelligent AI models US wants to keep for itself can be deployed here. In fact, it often seemed to me that Jay Powell was himself a superintelligent humanoid robot.

The only problem - the biggest one - would be 'independence', which is fundamental to the federal reserve system - in principle at least. While a totally human Fed has appointees pushing political agendas and personal idiosyncrasies / biases, and then there's an occasional law-suit on the Chairman, an agentic AI based system can be misled with false data. And who knows, agents may also have personalities and affiliations, depending on what went into their training. Which means for a well-intended agentic AI based Fed, there has to be a tightly controlled training mechanism, that ensures the agents are unbiased, transparent, goals-driven, meticulous, diligent and clean. But the interface with humans is where it'll all still fail, given humans will never let their control go, for both good and bad reasons. Good reasons - we can only trust humans to protect our best interests. Bad reasons - humans are divided at various levels - individual, communities based on religion, region, beliefs, experiences, economic status, affiliations, and so on; interests served vary, and not necessarily in any order or priority, and not generally good for most. What if you extend agentic AI to all governance so that Fed is not meddled directly by humans? One - AI can screw us more bigly and widely. Two - the interfaces with humans would still be there, albeit farther from the center of the agentic system, and humans at the interface can still manipulate AI to screw us. The mechanisms would be far too complex to comprehend and the outcomes far too unpredictable.

Seems humans anywhere are a problem after all. But then, who is AI for?

And...

Isn't AI 'human' too?

Fun question - What do you think an Agentic AI based Fed would have done with the policy rates yesterday?

How is human expression changing because of how LLMs express?

It's well recognized that LLMs have a certain writing style. Most of them follow common patterns like em dashes, 'a, b and c' structures, punch - then expand, 'it's x, rather than y', 'it's not p, it's q', and so on. But with so much content being written by LLMs, including that in research papers - where it's generally permitted to use AI for editing and improving readability, it's quite possible that the way humans write is also changing. With most of the content we consume being AI generated, the influence on our written expression is expected, and in my observation, starting to emerge. As evolved monkeys, most of our learning is through copying. Being quite intelligent, we are able to copy even complex patterns. The same intelligence allows us to adapt, improvise and customize - leading to both similarity and uniqueness in every aspect. (The previous line looks most definitely AI.)

I would expect the adjustment to be mutual, but I don't know whether AI is still learning from us in the way we do from it. But in future, we'll definitely have difficulty figuring out what's AI generated, because with time, NI would start looking like AI. (NI is natural intelligence. I guess I can (can I?) claim the term as the 'natural' counterpart to AI, which existed before AI.)

But the bigger concern is that, perhaps, we may not have any purely NI based writing in some distant future. We'll only be fixing drafts generated by AI based on what we tell them, just like most coders only tweak pre-written code. What we 'tell', and how we do, would be the only pure and natural expression from NI, although not polished to the eventual intended format. It's hard to imagine how that informal expression might change, but it certainly will.

There is still some way to go before we start talking to AI more than we talk to humans. That will have interesting ramifications on how we communicate, not only with AI, but also with other humans. (A close parallel, purely natural, is how American accent(s) came about, although nobody in the source countries of the OG immigrants in America spoke like that. Same goes for Australian accent.)

It's interesting how the various forms of human expression would evolve with AI playing an increasingly central role here on. I'll be watching this closely. And would love to hear what others are observing and experiencing.

IT services outsourcing - trapped in the cost play

IT services business started and evolved as a cost-reduction offering. First, offshoring offers a direct drop in costs to customers. Then greater productivity - doing more with less resources - offers greater cost savings with time. It's funny how most IT deal reviews by leaders hover around ON-OFF (onsite-offshore) ratio and YoY (year-on-year) productivity, while trying to hit an attractive price at an acceptable margin. Often, it's people with limited understanding of technology trying to buy and sell technology services.

These ratios, of course, are just high-level indicators of solution optimality, and benchmarks like 90-95% offshore headcount and 50-60% productivity-gains over 5 years are hard to push further. It's all in excel after all.

Seeking higher productivity has helped shape IT services with efficient processes, tools, and mechanisms, including automation and AI. But the pursuit is endless. Every new large deal, you are still asked - how will you bring your team to half its size in 5 years? There are no real, rather complete, answers. There's some experience, at times. There's automation. There are tools. There's AI. There's God. There's - 'we'll figure out when we get there'. Mostly, there's leap of faith compelled by fear: if we don't offer this, somebody else will, and the deal will be theirs; we have to take the risk... a calculated risk.

Peeling layers, you start looking for people mix - the pyramid. More the junior people, lower the cost - that's one lever you can pull. What about skill? - we'll train them - let's assume for now; and there's AI.

Once you've totally beaten this chain iteratively to its bare minimum cost configuration: # of people - ON-OFF ratio - Pyramid - Productivity - # of people - ON-OFF ratio - Pyramid - Productivity - ... You'd now turn to other costs - infrastructure, shifts, cabs, management, overheads, nickels, dimes, chillar!

For at least a decade now, IT service providers have recognized this need to get out of the cost play and move to value play. Mostly through talking, without really investing in genuine capabilities. So, outcomes have been only in pockets. The problem, I think, is in the nature of the buyer-supplier relationship - the relation is designed to be transactional, and both parties enter with short-term motivations.

And that probably explains why GCCs are emerging as a stronger value creation model.

The story around the picture...

Often when I look at a picture posted by someone, I can't stop myself from thinking how that moment came about, what happened before, what happened after, how that day must have looked like for people in the picture, how they interacted, what was going in each of the minds, what happened! They say 'a picture is worth a thousand words'. But the beauty of life is in what the picture doesn't convey but merely represents - all of which is in memories of those in the picture - each his/her own, fading slowly with time. And after long, the picture remains to serve only as the reminder that the moment truly happened.

Solution Modeling and Pricing Tools in IT Services Companies

I have been part of presales and solution architecting in various IT services firms. One common struggle and frustration in all these firms is with using tools (software, often web-based) for solution modeling and pricing. They keep trying, but just can't get them to work reliably.

The reason may be within the underlying data and the process. And it's complicated.

The sequence of steps to get to a price, in simple terms, is: resource-estimates → resource-loading → costing → pricing. In other words, there's a 'solution' to customer requirements - which is primarily the 'how', but also the 'who', the 'what', the 'when' and the 'where', which would determine the cost of doing it.

(1) Understanding the requirements, depth of solutioning and confidence in resource estimates is the foundation the commercials would be built on. This is also the first input, and the most common source of errors.

(2) The solution - including the estimates, staffing and the delivery model - have to be mapped into the modeling/pricing tool, coz the architect who built the solution is unlikely to be able to do it 'on' the tool; it's a technical and creative process.
On the surface it sounds like a simple porting of data. But this is the point where using the tool starts feeling like a waste of time, especially for large and complex deals. The transferring of data is pretty much manual, and each major or minor change in solution will need a repetition of the whole effort. A lot of tools have capabilities to read excel sheets, but those are rarely seamless, and still require manual effort to get to certain formats each time something changes.

(3) The next input tools require is access to an accurate, exhaustive, complete and well-maintained database of costs. Besides people, there are a whole host of delivery model considerations that determine a lot of the costs, most of which are very context specific. Most companies don't have such databases with all the characteristics listed above. They end up working with averages or ballparks - another source of errors.

(4) Solutioning and pricing are iterative, and may involve numerous rounds of back-and-forth adjustments, even manipulations, to make pricing attractive. Besides for the humungous manual effort it implies, there are things that don't add up, which are easier done in excel spreadsheets.

Errors add up in strange ways.

Conventional wisdom says processes and tools bring greater efficiency and reliable outcomes. But in the case of solution modeling and pricing, the tools seem to represent vicious circles companies don't know how to get out of, yet they are things they can't get rid of.

Companies that have the best handle on the four areas - (i) depth and confidence of solution, (ii) standardized solution modeling, (iii) robust cost database, and (iv) minimum solution meddling for better pricing - also have the most successful delivery, manage risks better, make healthier margins and have sustainable growth.

China to publish less in international scientific journals

Chinese scholars being urged to prioritize domestic journals as opposed to 'international' journals may open up some room. I wonder who'll take it. How wonderful it would be if the Indian context and Indian voice became so interesting to the international journals that they'd want more of it. Journals must reflect the choices of the audience. And audience for research requires not just a mindset, but an economic model where incentives from hyperinnovation are enormous, where institutions reward novelty, where perfection and aesthetics take precedence over fail-fast, and where capital is abundant. People in academia deliberate on the importance of having some top journals from India, with the highest reputation, standards and reach. But do we have the audience? Are they asking for it? Are they ready for it? We probably need the ecosystem first and the rest will follow.

China cools on overseas publication of scientific research

Originally posted on LinkedIn on 08 July 2026

Empathy, Emotional Intelligence - ‘show’ vs ‘have’

Past couple of decades, the discourse on personality, relationships and values has laid a lot of emphasis on empathy and emotional intelligence. While the underlying intent is goodness of heart, and the emphasis intends to condition humans to be good to each other and also to themselves, the outcomes we are marching towards are performative, rather than internalized. We are taught to ‘show’ empathy, not necessarily to ‘have’ empathy. We must ‘appear’ emotionally intelligent, not necessarily to ‘be’ emotionally intelligent. A key aspect to consider here is the vantage point, of course, i.e., the world sees what it sees about you — so “look” the part; and you know what you know about yourself — so “be” the part. The worldly material incentives are tied to performances. The internal incentives (which are?) are tied to being true to oneself. Through parallel discourses on personal branding, looks-maxing, and individualism, we are only validating performance as the means to worldly success, without offering any incentive to inculcate genuine human values. I wonder whether there’s truly a way to do the latter? And what kind of success would it offer? Why does it often take experience and maturity? What if we aren’t truly anything? What if we are all just performing — as best as we can and to the extent our abilities and impulses allow — to being whatever helps us get what we need, want, desire, crave or love?

Let's make some theoretical contributions today

Good morning. As I am sucking this 'heart' in, and starting my day, the thing I am thinking about is "Theory". Ability to theorize is what differentiates a scientist from a mathematician, a statistician, a data analyst or a consultant. While all of them draw 'conjectures' and rely on data to validate them, only a scientist 'hypothesizes', tests, and offers an explanation - a theory - one that works most of the time, preferably 99.9% of the time. Until it doesn't. Until someone, also a scientist, finds a nuance which demands an additional explanation, offers one, which becomes a 'theoretical contribution' - and then, with the padding, the theory claims to go back to 99.9% success rate.

Let me pour the 'heart' out and make some theoretical contributions today.

Originally posted on LinkedIn on 17 June 2026

From memorizing to now seeking the illusive originality

As I was making my daughter do her homework, I was reflecting on how our minds were trained as we grew up. Throughout primary school and even until the 10th standard, the predominant approach to 'studying' was to learn by-heart the questions & answers. One of my teachers often used the phrase "commit to memory, vomit on the paper". Exams were fundamentally memory tests. Writing as the textbook said was rewarded with highest marks, especially in subjects like social science, general science and languages. But in later education, especially in areas that involve academic writing, more so in research and publications, one is supposed to be extremely careful not to copy verbatim. Plagiarism is next to crime in academic ethics. Original writing carries value. But because we are trained to do everything by copying, we learn even original writing by copying, in a very ironic sense. Rather, we mimic - the style, the words and the broad structure - and fit in our thoughts. And now here's AI, and plagiarism doesn't matter anymore. Originality now has other dimensions, and we have opened all of them to adultery.

Originally posted on LinkedIn on 15 June 2026.

em dash—but only when you need it

The constant urge to post on social media, driven by wanting to be visible and relevant all the time, has become the defining characteristic of a certain stratum of people in our times. It offers a feeling of being watched, performing and contributing. And in the process—right from personalities on the top to aspirational life-coaches—people are throwing around garbage or in-process uncertain, unshaped fluff that the world would be better off not consuming until it has gained some genuine ground and has reliable informational value. It could be done for signalling. But a signal is effective only if the intended target is clear and the chosen medium of transmission aligns with the target with significant precision, so that there’s minimal unintended loss. For precise targeting, you can’t use a dispersed medium. And if a dispersed medium is used, then probably you don’t intend precision. You intend collective and widespread phenomena like mass-opinion, chaos, panic, doubt, uncertainty, or divided opinion amplification. This behaviour manifests everywhere: from our workplaces to WhatsApp groups to other social settings.

Anyhow, I must point out: the em dash above has come from my own keystrokes! I didn’t know of the existence of this character nor notice it as distinct until AI pushed it on the world relentlessly. And it took me some time to figure out how to type it. It is strangely satisfying that I can do what AI does.

The beauty of being one with your story

Many years ago, I tried to write a novel. It was ambitious of me, driven by the belief that “I can also do it” after reading Chetan Bhagat, yet without a good understanding of the effort it takes. I tried to weave a story, combining and building on a few bits and pieces of what I had written on my blog. Those pieces involved a fictional character called LaajVanti, and were casual, funny and relatable for readers with my kind of profile at the time. I tried to write the novel like we do a course assignment or corporate proposals - put together some nice content and write stuff connecting them all - stitch them together, ensure a flow, cook up a storyline and conclude like it conveys beauty.

In corporates we often start with a strawman when there is a boss to review, time to spare, and the broad structure is unambiguous. Writing fiction, however, is a creative process and cannot be forced to follow a set structure. It involves imagination that can suck you into itself, such that you coexist in various contexts with your characters - either being one of them or watching them, standing next to them. It consumes you so much that you literally live in many imaginary worlds for extended period of time, and this physical world just becomes a pathway to those. These other lives in other worlds bring strong emotions - curiosity, intrigue, anger, excitement, joy, melancholy, hope - in forms and depths you rarely experience in real life. Getting there, however, is not so spontaneous. It requires commitment, persistence and internal push to cross some boundaries. Until you cross those boundaries, it’s a struggle, a hustle, some trial-and-error, and mostly shooting in the dark to see sparks and patterns. It requires patience. Lots of patience. You let your mind wander. You let it dream. You’d often wake up without finishing a dream, especially when it was taking an interesting turn. You’d feel disappointed at the broken sleep and try to push yourself back into it, just so you could get to a conclusion in the dream. Sometimes when your sleep is shallow, you mix dreams with your own cooked up details (hallucinations?). And then, you can only sleep so much at a time - a limitation that frustrates you. And not all of it is about your story. But then you push harder, dream harder, free your mind, let loose your imagination, and then you don’t need sleep to be in peace, in dream or in any world you choose to be. You become one with your story.

I experienced traces of it, but I couldn’t cross those boundaries. I told myself that fiction isn’t my thing. I may have lied; I don’t know. I’ll try again.

It is. Is it?

I heard Yann LeCunn recently say, and I am paraphrasing, that for AI to be truly meaningful, it must have the ability to predict outcomes of its actions and adjust itself accordingly. Use of AI for reasoning - work with us to brainstorm, build on ideas - is one of the most meaningful uses of AI. However, often times, AI ends up becoming an echo chamber without us realizing it. It can make your noise sound like music. You may end up loving your voice, singing louder and generating more noise, until the audience tells you they can't bear it.

Working with AI to test and build ideas can often take you too far down wrong paths. Most of these LLMs, if not all, are very good at offering reinforcement to your thought process, adding layers of rationalization, generating confident and seemingly valid grounds for various elements of solution options that you piece together, and then shaping for you the artifacts that you feel proud for having created - and your LLM doesn’t mince words in making you feel so. And yet, if you started with some wrong premise or assumption that was central to the whole pursuit, and built everything on top of it, the process would only lead you farther away from any meaningful solution to the problem you set out to solve.

Posted on LinkedIn on 02 June 2026.

AI as a 'resource' for Research

Impact of AI on research is the most talked about topic in most academic circles and research conferences these days. In one such discussion during the India Strategy Conference (ISC2025) at Indian School of Business this year, I raised a concern that if most research gets 'powered' by AI - which is increasingly the case - the quality of research outcomes will be increasingly dependent on the kind and version of AI one has employed. As AI becomes a resource, ability to carry out high quality research becomes a question of access and affordability.

I was countered with an argument that AI is an equalizer. In one sense it is, as it allows one to try his/her hands at high quality research without the depth of skillset that research demanded earlier. But beyond the basic level, inequality starts taking shape because the resource is limited and controlled. Claude Haiku 4.5/Sonnet 4.6/Opus 4.7? ChatGPT Free/Go/Plus? Gemini Free/Plus/Pro/Ultra? We all know that the free ones are extremely limited in capability and reliability. Paid ones have variants offering different levels of capability depending on how much you pay. The pricing is equal worldwide, but the purchasing power isn't. So users are forced to settle, and work with whatever is the best they can get their hands on.

Research outcomes have always been constrained by access to resources. But now, with 'intelligence' as a resource, the race can get too exhausting for many - some running on sand, some even on water - while a few cruising on firm surfaces, breezing through in skates that don't even make a sound. The bodies may be equal, the race still isn't.

Posted on LinkedIn on 27 May 2026.

Claude and its usage limits

I appreciate the makers of Claude for being so thoughtful and responsible to have built ‘work-life balance’ into it. And neatly optimized it for the $20 subscription - it allows the right amount of usage that exhausts - not too soon, not too late - then making you wait to watch reels, until you can resume using intelligence - the artificial one.

Originally posted on LinkedIn on 25 May 2026.

Short-termism is Sticky - Moving from Principal-Agent to Principal-Principal Conflicts

Now this is the other side of the corporate short-termism debate that has gained little importance. If QoQ disclosures to shareholders is forcing companies to think short-term, moving to a less frequent disclosure can create information asymmetries that can harm retail investors / minority shareholders who rely only on public disclosures and media with wide reach. The measure intended to alleviate short-termism driven by principal-agent dynamics is probably going to amplify short-termism driven by principal-principal conflicts!

I recommend reading the full letter below from wallstreetbets community to SEC!

Originally posted on LinkedIn on 16 May 2026

I wonder...

I am sipping coffee at Starbucks, trying to work, and ended up 'wondering' - May be it's not AI that's making us useless... May be it's LinkedIn. It has conditioned a whole bunch of people to start 'wondering' after every little thing they do, or don't, or just go through, or don't... An hour of wondering per minute of activity - that might as well be wondering - then an hour trying to post on LinkedIn what we wondered, often by getting AI to write about it beautifully... Then a few hours checking likes and comments, wondering about those. We even wonder about wondering. Many are wondering - even worrying - whether they are wondering enough. People have started wondering together. There are even courses to help you do high quality wondering. A cursory scan of LinkedIn will tell you that we are certainly getting better at it. AI is playing a huge role in wondering lifecycle management (WLM). It's even executing some of the steps in the workflow. Can it do the whole of it? I wonder. Soon, perhaps. But we should make sure we are always wondering 10 times better than AI. A lot of us post our pics to go along with our wondering narrative. AI can't. But then it can generate a prettier human picture than any of our wondering faces - I wonder whether that would matter. I wonder whether hallucinating by AI qualifies as wondering of some sort. Or is it just a couple of steps behind wondering? Should we worry about it, I wonder. We have to outwonder AI, as wondering is all we got left. I wonder whether we can do it. We have to. It's our 'wonderful' world afterall! But then, I wonder...

Originally posted on LinkedIn on 12 May 2026