fact-checking on S Anand

My predictions in 2025

Mon, 17 Aug 2026 09:20:33 +0530

In 2025, I made a number of predictions on this blog. (Not intentionally. I just said stuff.)

I asked ChatGPT to audit them. It selected 440 claims, filtered out vague or pending ones, and verified the rest. Here’s what I got right and wrong.

🟢 “My chat will overtake search in 12-18 months. When ChatGPT becomes my primary lens on knowledge…” (chatgpt-vs-google-usage.md:L27). Audit: This actually happened in your own browsing data: search led through April 2026; in May, chat jumped to 2,211 visits vs 1,319 search visits, and stayed comfortably ahead thereafter.
🟢 “Typed languages are better suited for vibe coding. This will likely lead to the growth of typed languages (TypeScript, Rust, Go) but also of typing in untyped languages (e.g. Python).” (things-i-learned-10-aug-2025.md:L33). Audit: TypeScript became GitHub’s #1 language in August 2025 and grew 66% YoY; GitHub itself explicitly connects the rise of typed languages to more reliable AI-assisted coding. This is unusually strong because you got both the direction and mechanism. (The GitHub Blog)
🟢 “Code Mode … is a smart way to use MCPs and a very likely future direction. Using LLMs to write code to call MCPs rather than directly.” (things-i-learned-05-oct-2025.md:L49). Audit: OpenAI’s current Responses API has essentially this as a named capability: Programmatic Tool Calling lets a model write and execute programs that coordinate multiple tools and intermediate results. (OpenAI)
🟢 “CLI optimization for LLMs will likely emerge. More CLIs (and wrappers / hooks in the shell) will improve output and error contexts for LLMs…” (things-i-learned-27-jul-2025.md:L86). Audit: By May 2026 you were yourself testing rtk, a CLI proxy explicitly producing compact agent-friendly command output; across 216 commands you measured about 50% token reduction. That is almost exactly the wrapper you predicted.
🟢 “In the future, AI that works directly with file systems, Model Context Protocols, and local APIs are likely to become more important.” (features-actually-used-in-an-llm-playground.md:L76). Audit: File-system-native coding agents became mainstream, while MCP expanded into both hosted and local integrations; Anthropic now packages local MCP servers as easy-to-install desktop extensions. (Claude Help Center)
🟢 “Agents are slow. Parallelizable tools … will grow. Tool speed … will become more important.” (things-i-learned-22-jun-2025.md:L106). Audit: Parallelism has become a central agent UX: OpenAI’s Codex app is explicitly built around managing multiple agents simultaneously; at the extreme, internal users now accumulate more than 60 hours of agent turns per day by parallel execution. (OpenAI)
🟢 “Companies-of-one will grow. Sole founder can handle support functions.” (things-i-learned-24-aug-2025.md:L14). Audit: Nasdaq’s Economic Institute finds one-person US business applications up more than 20% since early 2025, with essentially all recent application growth coming from solo businesses. Stripe separately reports solo founders reaching 63% of Atlas C-corps in Q2 2026. (Nasdaq)
🟢 “We will move towards an organization structure where developers are embedded with business teams rather than working as a separate group. Sort of like embedded executive assistance instead of a central typing pool.” (things-i-learned-08-jun-2025.md:L49). Audit: Forward-deployed engineer demand reportedly increased 42-fold from 2023-25, with roughly 9,000 roles globally by early 2026. The role is almost precisely “technical people embedded with the business to make AI work in its real environment.” (Reuters)
🟢 “Shadow apps will grow. Anyone can code. Users build apps with prompts, sheets, agents, outside of IT SDLC. Like Excel sheets.” (things-i-learned-24-aug-2025.md:L18). Audit: Microsoft now explicitly describes a “new wave of shadow AI”: users installing coding/desktop/SaaS agents outside traditional IT governance, and has built discovery products specifically for unmanaged AI applications and agents. (Microsoft)
🟡 “Agents generate diffs/PRs. Tools to edit and comment on these online will emerge.” (things-i-learned-22-jun-2025.md:L107). Audit: GitHub now measures PRs created and merged by Copilot coding agent, while review comments can be handed directly to the agent with “Fix with Copilot,” including batches of review feedback. That’s almost verbatim fulfillment. (The GitHub Blog)
🟡 “Models’ ability to orchestrate longer workflows will improve. Factor that into your application design.” (things-i-learned-10-aug-2025.md:L44). Audit: By mid-2026, OpenAI reports large increases in requests corresponding to >30-minute, >1-hour and even >8-hour human tasks, while Codex explicitly targets long-running tasks spanning hours or longer. (OpenAI)
🟡 “Code review process will be re-invented.” (things-i-learned-22-jun-2025.md:L109). Audit: GitHub has rebuilt Copilot review around an agentic architecture that gathers broader repository context, uses tools, produces findings, and can hand fixes to another coding agent. This is substantially more than autocomplete added to old review. (The GitHub Blog)
🟡 “Domain expertise will therefore become even more valuable in the near future.” (things-i-learned-20-apr-2025.md:L39). Audit: 2026 hiring evidence points toward domain/product expertise becoming more important rather than pure coding alone, particularly as AI handles more implementation and firms need people who can connect it to actual business functions. (Reuters)
🟡 “Validation is the New Bottleneck: Since coding is now much faster, the critical, time-consuming task has shifted to reviewing, testing, and validating the LLM’s output.” (things-i-learned-17-aug-2025.md:L90); you also predicted “The Quality Control (QC) function will become larger and more critical” (L95). Audit: GitHub has now productized exactly that bottleneck in Code Quality; more than 10,000 enterprises used its preview, and GitHub explicitly frames AI-accelerated code output as creating the need for trustworthy pre-merge quality validation. (The GitHub Blog)
🟡 “Agents generate technical debt faster than humans. Solving this will become a major problem/opportunity.” (things-i-learned-22-jun-2025.md:L114). Audit: GitHub’s 2026 Code Quality launch is close to a commercial instantiation of this forecast: AI increases code output, so automated quality/debt detection and remediation moves earlier into the development cycle. (The GitHub Blog)
🟡 “Governance will grow. Non-experts are acting like experts. Validation is more important.” (things-i-learned-24-aug-2025.md:L19). Audit: The companion to shadow AI has indeed been governance: Microsoft now ships specific discovery, monitoring and governance for unmanaged AI agents, while NIST has continued expanding formal GenAI evaluation tooling. (Microsoft Learn)
🟡 “Soon, we won’t just follow a lesson plan – we’ll have lessons built just for us. AI will track how we learn and adapt in real time. It’ll feel like having a personal coach in your back pocket.” (o3-is-now-my-personalized-learning-coach.md:L91). Audit: ChatGPT Study Mode now asks what the learner knows, adapts explanations, checks understanding, works from uploaded course material, and uses memory to personalize support; OpenAI explicitly describes the objective as personalized learning support available to any student. (OpenAI Help Center)
🟡 “Cost is going down so quickly right now that all you have to do is wait, and stuff will become available for a very affordable or even a free price.” (things-i-learned-16-mar-2025.md:L116). Audit: The broad direction held. OpenAI cut GPT-5.6 Luna API prices by 80% in July 2026 while simultaneously improving capability-per-dollar. The “all you have to do” part is hyperbole, but the price-curve forecast was right. (OpenAI)
🟡 relayed: “Control of chips and GPU compute is what will likely be the gameplay to control AI dominance globally.” (things-i-learned-02-feb-2025.md:L16, attributed there to Dario Amodei). Audit: Advanced-AI-chip export licensing remains an explicit geopolitical control mechanism in 2026, including restrictions and license review for H200/MI325X-class accelerators going to China.
🔴 “AI closes the gap between junior & senior devs – even when both use AI. Quality doesn’t suffer much. So onboarding can be faster, compensation ladder may shorten.” (things-i-learned-03-aug-2025.md:L52). Audit: The emerging evidence says AI changes the work but does not erase the expertise gap: experienced developers are better at steering/delegation, while low-experience AI-heavy contributions incur substantially more review and lower acceptance. (arXiv)
🔴 “LLMs already deliver hours of analyst work in minutes. Entry-level roles WILL vanish.” (goodbye-mba-hello-ai.md:L17). Audit: The labor-market warning was directionally good, but “vanish” is a major magnitude error. Stanford finds a meaningful relative decline among 22-25-year-olds in highly AI-exposed jobs, while employment remains substantial and overall exposure groups still show employment growth. “Entry-level hiring contracts sharply” would have scored much better. (Stanford Digital Economy Lab)
🔴 “Coders micro-manage LLMs. I think a novice will be more efficient and get better results than me.” (how-to-visualize-data-stories-with-ai-lessons.md:L285). Audit: Current empirical work points the other way in real software work. In a 22,953-PR study, lower-experience AI-heavy developers received 4.5* more review comments, had 31% lower acceptance, and took over 5* longer to resolve issues; qualitative work likewise finds experienced developers better at delegation and control. (arXiv)
🔴 relayed: “API access from model providers will shrink. Selling tokens is not a viable business model given lowering costs.” (things-i-learned-23-mar-2025.md:L19, from the Alexander Doria notes immediately above it). Audit: Almost exactly backwards. Model providers expanded their APIs into richer agent platforms, and token-metered API access remains a core commercial model - including premium pay-as-you-go modes. (OpenAI)
🔴 “APIs are likely to be replaced by just chat requests that will do the same thing. APIs might be replaced by RPA, where somebody uses a chatbot to do the equivalence instead.” (things-i-learned-16-mar-2025.md:L111-L112). Audit: Chat did become a front end, but the implementation moved toward more APIs underneath, not fewer: tool APIs, Responses, MCP, computer-use interfaces and programmatic tool calling are now the substrate agents use. (OpenAI)
🔴 “Software companies build ‘SaaS’-like apps today. Agents will replace apps. Instead of UI, workflows, and app logic, they’ll engineer prompts, APIs, and evals.” (agents-will-replace-saas-apps.md:L12). Audit: The interface-shift was right; “replace” was not. Gartner now forecasts agentic AI may expose roughly 20% of SaaS application spending by 2030 - meaning substantial disruption, not app extinction. Agents are often a new interaction layer over systems of record and APIs. (Gartner)
🔴 relayed: Models will “internalis[e] workflows … to wipe out the apps and workflow space.” (things-i-learned-23-mar-2025.md:L17, from Alexander Doria notes). Audit: “Internalize capabilities” was insightful; “wipe out” was the failed extrapolation. Enterprise applications remain a very large substrate even in Gartner’s fairly aggressive agentic-AI forecast. (Gartner)
🔴 “Demand for SaaS (one-size-fits-all) will shrink.” (things-i-learned-06-apr-2025.md:L89). Audit: Not yet. For example, Gartner forecasts Indian SaaS spending growing 18.9% in 2026, from $3.9B to $4.6B. AI is changing SaaS economics and seat licensing, but current demand is still growing rather than shrinking. (Gartner)
🔴 “The early majority have come in… Soon the late majority will come in asking for existing solutions that have already solved their problem for many others.” (things-i-learned-03-aug-2025.md:L44). Audit: This mapped your client/audience experience onto population adoption much too quickly. In 2026, US Census data put business AI usage around 17-20%, nowhere near a conventional late-majority phase. (Census.gov)
🔴- baseline error: “Given the cost and accessibility of drones, I guess drone terrorist attacks will soon emerge.” (things-i-learned-16-nov-2025.md:L43). Audit: They had already emerged. The UK government documented Daesh using small armed remotely piloted aircraft carrying grenades in Iraq in 2017; the UN had already been studying weaponized UAS use by non-state armed groups for terrorism-related purposes before this 2025 post. This is therefore a clean failure to establish the baseline, not a future hit. (GOV.UK)
🔴 relayed: “Personal writing with connection won’t go away. AI can’t give you heartbreak. But the rest of non fiction writing will vanish.” (things-i-learned-30-mar-2025.md:L69, under “Notes from Writing with AI”). Audit: Nonfiction is under real pressure, but “vanish” is nowhere close. UK nonfiction publishing still generated about GBP1.0B in 2025, down only 3%; the wider publishing industry reached record revenue. (Publishers Association)

Legend:

🟢 = I got it right
🟡 = I might be right?
🔴 = I got it wrong

I was wrong when making extreme predictions, e.g. APIs, SaaS, workflows, junior devs will be replaced by agents. They did shrink (so I was directionally right) but they’re far from replaced.

I was right when making directional predictions, e.g. CLIs, shadow apps, verification, solo founders, … will grow. These are easier and safer to predict (hence less impressive).

There’s a meta-insight here. You can take anyone’s corpus (meeting transcripts, emails, blog, books, research) and see how good their predictions are. This makes auditability for more accessible. So, I predict that we’ll see new roles & business opportunities emerge in 2026 based on this. For example:

Auditing researchers’ predictions
Auditing analysts and industry experts’ claims
Auditing company forecasts and promises
Auditing leaders’ promises
etc.

How IMF mis-forecasts GDP growth

Mon, 22 Jun 2026 10:41:30 +0800

The IMF forecasts GDP growth every year. Their forecasts for the current year are slightly low. Their forecasts for the next year are slightly high. After that, it remains high.

Some forecasts, like China, Singapore, UAE, Equatorial Guinea are consistently low.

Other forecasts, like Japan, Congo, Mexico, Pakistan are consistently high.

The interesting meta-pattern is how this sort of past-forecast analysis can be done for any topic. This emerged from an Ethan Mollick post and then I asked:

Me: Ethan Mollick had GPT-5.5 Pro update his grad school paper. Cool! What can I do like this?
Claude: Analyze your past work - like your blog posts.
1. Fact-check it
2. Find wrong sources
3. Critique it
4. Find contradictions
5. Find buried ledes / unbuilt sequels
6. Find what aged badly. What’s the pattern?
Me: Um, what about others’ work?
Claude: Sure!
1. Check IMF GDP forecast accuracy
2. Check target market size forecasts
3. Check company / asset price forecasts
4. Find which books are based on studies no longer true
5. Find sources of popular stats that are wrong
6. Find what numbers people quietly revised
Me: Where can I find IMF GDP forecasts?
ChatGPT: On the IMF Data WEO site
Me: Check IMF GDP forecast accuracy.
ChatGPT: The IMF has a modest optimistic bias for future-year GDP-growth forecasts, but not for same-year nowcasts.
Me: Translate that to English.
ChatGPT: For the current year, IMF typically underestimates. Next year onwards, they typically overestimate.
Me: Why? How?
ChatGPT: IMF has systematic biases.
- Advanced economies are easier to forecast and slightly less biased
- Some countries (China, UAE, …) are largely under-forecasted. Others (DR Congo, Haiti, …) are consistently over-forecasted.
- They underpredicts extremes in both directions.
- When their forecast is very high, it is more likely to be too high.
- Corrections are timid. They don’t close old error gap until very late.
Me: Draw this as a dot chart.
ChatGPT: OK
Me: Spread out the dots.
ChatGPT: OK
Me: Let me filter by country.
ChatGPT: OK
Me: Let me permalink to the selection.
ChatGPT: OK
Me: Split this into a HTML and JSON file.
ChatGPT: OK

… and here is the data visualization.

Add a Verify Button

Sat, 30 May 2026 11:39:10 +0800

Rohit Saran looked at the Statoistics cards my AI agents are generating for The Times of India, and asked about a small button under each one.

In the list of Statoistics that you had put, I saw there’s a button called ‘Verify.’ What was that meant to be or will do in future?

That verify button explains the claim, mentions the sources, and shows how to check the claim.

One card said “9 in 10 Indians want a family doctor and barely 1 in 35 has one”. The button breaks that down:

“87% want a family doctor, 2.8% outpatient visits were to an Asha worker…”
It identifies in the source document what are the columns that we were looking at, what numbers it verified.
It links to the program that it wrote to do the verification.

I said, “it lets humans check if the numbers are right - by giving them steps – where exactly to check, how to check if it is correct.”

Sajeev pushed back: “It’s more ’explain’ than ‘verify’ really.”

True. Saurabh had asked for exactly this earlier: while a person is checking by hand, give them something that shows how the AI got to its answer. A verify button’s first job is not to prove the AI is right. It’s to let a nervous journalist check, cheaply, until they stop being nervous.

This instinct is old. The Royal Society took nullius in verba as its motto around 1662, “take nobody’s word for it.” They didn’t print claims and ask you to trust the author. In 1663 they made Robert Hooke their Curator of Experiments, whose job was to re-run the demonstration in front of the Fellows. A verify button is that, without Hooke.

(Merchants got there two centuries earlier: double-entry bookkeeping, codified by Pacioli in 1494, means every entry has a counter-entry and the books either balance or they don’t.)

Rohit’s reason for liking it went somewhere I hadn’t fully thought through. He went to brand.

It’s like why a product with 10-year guarantee is likely to be made better than a product with 2-year warranty, because the company has confidence to tell the customer, ‘Look, I am standing behind this product for 10 years.’

And later:

Any brand that is saying, ‘Whatever I write is verifiable,’ is so much more in this age of misinformation.

His version of why this matters for a newspaper: “a brand is only about trust. Rest is news is anyway a commodity.” A verify button is a public claim that you’re willing to be checked.

Here’s how I actually build one “Verify” buttons, in increasing order of effort.

Link plus a searchable string. A hyperlink may still be wrong. I want a link and a short quote I can paste into the page’s search box and find. “When I click on that link, I should be able to literally search for and find that piece of text, verifying that it did not hallucinate” Then even a plain program (not even an LLM) can open every link and confirm the text is there.
For numbers, the SQL query. If it’s data, the SQL query (or Python script) that fetches that particular result is the closest equivalent. The button should just run the query against live data and shows the number. The user doesn’t need to know SQL - they just see that the number matches.
The procedure as a checklist. The button breaks the card into steps: this is the claim, this is the number, this is the column it came from, check that the D1A value matches. A person ticks down it.
Verify with an AI agent. Add a link that opens the claim in Google AI mode with a pre-filled prompt asking it to fact-check the claim. For example: Fact-check with step-by-step evidence: According to Citizen Survey 2022-23, 87% of Indians want a dedicated family doctor but only 2.8% actually use one. How might it have changed since the publication?

Rohit framed verification as three jobs, not one: “Verification has sourcing, verification, and updation.” The last clause lets you also ask whether the number has gone stale since you published it.

Getting the source right is not the same as getting the conclusion right. Rohit said: “you are asking AI not only to get right source and right data, but now we are asking to interpret.” And interpretation is subjective on both ends. The button can confirm the number is real but not prove the argument is sound.

Of course, the sources could be wrong. “Check the source” assumes good data quality. Luckily, data is more often right than wrong, and verification can shine a light on bad data.

We can start simple. The cheapest version: every AI output has a “Verify” link to a search query the user can easily inspect. That changes their question from “can I trust this?” to “let me check.”

If this can establish trust and a brand for India’s largest newspaper, enterprises AI apps might do well to follow.

Panchayat solves the wrong problem

Sun, 26 Apr 2026 19:02:18 -0400

In Panchayat Season 1 Episode 7 Ladka Tez Hai Lekin…, at around 17:00, Pradhan asks Abhishek to solve problem 42.

42. A takes 5 days more than B to do a certain job and 9 days more than C.
    A and B together can do the job in the same time as C.
    How many days would A take to do it?

    (a) 16 days
    (b) 18 days
    (c) 15 days
    (d) 20 days

The correct answer is (c) 15 days. But interestingly, ChatGPT got it wrong the first time too. It said (a) 15 days instead of (c) 15 days, and required a fact-check to correct itself.

But Abhishek’s steps have nothing to do with the problem - he proceeds to solve a completely different problem and ends up with the supposedly correct answer of 37/23.


(42)

Qty of x in container (100 l) = 40 lts
Qty of y in the container of 100 ltr = 60 ltr

Qty of Milk in final Mix = 100 × 3/4 = 75 ltr
Qty of water in final Mix = 100 − 75 = 25 ltr

[crossed / scratched working]

x = 40 × 2/3 , 40 × 1/3 = 80/3 , 40/3   [circled]

Qty of Milk & water in 60 ltr of Mix

y = (60 × a)/(a + 2) , (60 × 2)/(a + 2)
  = 60a/(a + 2) , 120/(a + 2)

Now, the problem I’m trying to solve is, which of these made the bigger mistake:

Abhishek, who solved a different problem
ChatGPT, who got the answer right but the option wrong
The writers of the show, who failed to correct this obvious error

How to use AI for research

Sat, 04 Apr 2026 23:30:53 +0800

I asked ChatGPT to research universities’ AI policies. Here is the report

Here are the four lessons I learned from that - about how to use AI for research.

1. Show examples of failures to avoid. Jivraj’s earlier research kept surfacing AI policies universities had researched, not written for themselves!. So I told ChatGPT to:

… double-check that they ARE, in fact, about their own use of AI - not policies they’re proposing for others or are researching.

This is called pre-specifying exclusions. Giving negative examples help. Wei (2022).

2a. “Double-check” doesn’t always work. Though I told ChatGPT to “double-check”, it still got things wrong. For example, it cited MIT’s AI policy home page as evidence that the policy covers students and faculty, just because the words were present. That’s not right!

Models get over-confident - and that’s exactly when they don’t double-check. Asking them to double-check is a good habit, but not fail-safe. Kadavath (2022)

2b. Expicitly tell it to find mistakes. I told it to:

Find mistakes in as many claims as you can.

This is stronger than “double-check”. It turns the model against itself, and it worked quite well.

1. Show examples of failures to avoid. (Repeat.) When asking it to find mistakes, I gave it the same example.

… MIT, “covers_faculty_or_staff” cites “quote”: “Students • Faculty and Staff • Visitors and Guests • Generative AI use at MIT”. But that’s actually a set of links to Students, Faculty and Staff, etc. It’s not evidence that the POLICY covers them - and I’m quite sure the policy isn’t for guests!

That’s few-shot prompting. Concrete examples beat abstract instructions.

3. Ask it to list failures explicitly. I told it:

I am also interested in universities that conspicuously lack a policy …

Without that, it might have returned only positive hits. Missing evidence and failures are important data, too!

4. Break large tasks into batches. When I asked it to research 20 universities, it made several mistakes. Instead:

This may be a complex task, so let’s just do this for the first four Universities.

Now, it didn’t make any mistakes! Sometimes, it gets lost in the middle for long tasks.

So there it is - the four rules of AI research I learned from this exercise:

Show examples of failures to avoid
“Double-check” doesn’t always work. Expicitly tell it to find mistakes
Ask it to list failures explicitly.
Break large tasks into batches.

Verifying Textbook Facts

Fri, 23 Jan 2026 13:36:50 +0530

Using LLMs to find errors is fairly hallucination-proof.

If they mess up, it’s just wasted effort. If they don’t, they’ve uncovered a major problem!

Varun fact-checked Themes in Indian History, the official NCERT Class 12 textbook.

Page-by-page, he asked Gemini to:

Extract each claim. E.g. “Clay was locally available to the Harappans” on page 12.
Search online for the claim. E.g. ASI site description and by Encyclopedia Britannica.
Fact-check each claim. E.g. “Clay was locally available to the Harappans” is confirmed by both sources.

Here is his analysis and verifier code.

Before hitting rate limits, he scanned 56 pages and fact checked 45 claims from a dozen pages (detailed report). Even among this small sample, there was an interesting error on Page 22:

Only broken or useless objects would have been thrown away.

In reality, migrants abandon useful objects and ritual deposits give us intact, useful objects.

So, is this an error in the textbook worth correcting? Maybe. Worth exploring.

BTW, the original analysis hallucinated sources: “Discard and Reuse in the Ancient Near East” and “Recycling in the Bronze Age” from Cambridge. But for fact-checking, this doesn’t matter. The point is to flag questionable claims for further review.

There are others:

Did R.E.M. Wheeler take over as Director-General of the Archeological Survey of India in 1944 (according to the book) or in 1940 (according to other sources)? Worth checking the ASI records?
[Ashoka’s army was] 600,000 foot-soldiers, 30,000 cavalry and 9,000 elephants? The book says “some historians consider these accounts to be exaggerated” but the consensus is that these are almost certainly exaggerated. Worth rewording?

If LLMs find half a dozen questionable claims from a dozen pages, fact-checking textbooks seems an effective use for LLMs.

Especially when 25 million students read them every year.

I have personally suffered from errors in history textbooks.

I wrote my class 9 history exam in 1989-90. The Hindu published a series titled “This Day That Age” with news from 50 years ago. That morning’s was that Germany invaded Poland with Russian aid. I proudly wrote this in my exam and lost a mark, since the history textbook clearly mentioned that Russia opposed Germany.

(That I feel indignant 36 years later tells you how sorry my life is.)

New ways of reading books

Tue, 06 Jan 2026 11:03:07 +0800

I’m using AI to read books by:

Summarizing. This tells me what the books is about, the key points it makes and the main takeaways. It also helps me decide if I want to dig deeper.
Fact-checking. I can find mistakes, alternate perspectives, and biases. That’s a huge win!
Re-authoring. I can write it in the style of Malcolm Gladwell, Randall Munroe, Richard Feynman, or anyone else I like. Makes dense prose much more enjoyable.

So far, I’ve applied this at different levels - and I’m sure there are more possibilities:

A book. I can summarize & fact-check books like How Not To Die and learn what’s really established.
A chapter. I can turn each chapter of Primo Levi’s The Periodic Table into a Randall Munroe article that makes me laugh.
A sentence. I can take a single proverb like 忙しいほど幸せ or a Kural like அடக்கம் அமரருள் உய்க்கும் and turn it into an essay that makes me cry.
Multiple books. I can compare and fact-check multiple books on the same topic, e.g. Make It Stick, A Mind for Numbers, Ultralearning, and How to Take Smart Notes.

But I suspect this is just the beginning. AI opens up many more reading styles. For example:

Reviews over time. Could we read Lolita along with the 1950s interpretation, the 1970s feminist critique, the postcolonial turn, the contemporary reassessment? Understanding how meaning evolves over time?
Character story. Reading just the sub-novel of a single character, e.g. Hoid across the Cosmere?
AMA with the author. Interpret what the author might have meant or said, e.g. What caused the quake in Arelon?

We should be mindful that artistry is not understanding. Assessing, teaching, applying, debating, … these are better tests. But accessibility, enjoyment, and novelty are meaningful too. Let’s go exploring!

I count AI summarized books as "Read"

Sat, 27 Dec 2025 02:54:41 +0000

I have this nagging feeling (maybe you do too?) that it’s cheating and I’m not really learning if it’s so easy. The same voice makes me feel guilty when using coding agents to code or ChatGPT in meetings.

I’m telling that voice to relax.

I upload books to Claude and ask it to “Comprehensively and engagingly summarize and fact-check, writing in Malcolm Gladwell’s style, the book …”. I can read it in an hour instead of twelve. Four bullet points instead of forty. With (this surprised me) roughly the same number of insights I actually do something with.

I read books for pleasure, learning, and implementing.

That last bit is key, but the bottleneck isn’t the book. It’s me: how much I can absorb, retain, and act on. From what I can tell, between reading 300 pages vs 10, there’s not much difference.

The pleasure is intact

I worried I’d lose the emotional jolt – those mind-blown moments that make reading feel worth it. For science and self-help books, I haven’t.

For example, take this summary excerpt of How Minds Change

The irony is almost too perfect. To change someone’s mind, you have to stop trying to change their mind. You have to start by trying to understand it.
And when you do that—really do that—something unexpected happens. Sometimes their mind changes. And sometimes, if you’re honest with yourself, yours does too.

That blew my mind for a good 10 minutes.

Or, from Ants:

Think of a leafcutter ant colony as a high-tech Fungus Farm. There are four main players in this system: two are partners, one is a bodyguard, and one is a villain.

I loved the villain part!

There’s a learning bonus: fact-checking

When I ask Claude to fact-check the book while summarizing, I learn beyond the book. For example, I learnt that, among these longevity books:

Lifespan is largely rubbish.
The Telomere Effect has a decent core but builds fancy castles on shaky foundations.
Outlive is more careful with its claims, though less ambitious.

I learned this during the reading, not months later when some podcast corrected me.

You can also just… ask questions.

While reading Ants, “OK, aren’t most species superorganisms then? If yes, what’s the big deal?”
While reading The Cancer Code, “Why are free radicals produced? Give me examples. Are they always oxygen molecules that lack electrons? ELI15.”

The book doesn’t do that. The summary can. (I’m loving the ELI15 responses, BTW.)

I can read inaccessible books

I can now read Kamba Ramayanam in Tamil (a script I read quite slowly). Or Gitanjali, The Iliad, Dream of the Red Chamber, or more.

It’s not just language. I can read boring or intimidating books. On Growth and Form. The Road to Reality. The Structure of Evolutionary Theory. Brilliant but unapproachable.

Heck, I managed to read Ants after 30 years!

Sometimes the very first line of a summary hooks me so hard I can’t stop reading. That’s a strange thing to admit about a summary. But there it is.

I haven’t figured it out fully

I would read Brandon Sanderson or Jeffrey Archer this way yet. The point is the experience itself: the slow unfolding, the surprise, the voice in your ear. They write in a style I like. Why summarize that?

Biographies are trickier. The Choice’s summary made me cry. So, I do want to read The Choice in original - and the summary helped me prioritize it.

Histories? I’m not sure. Manga? No way.

I’m logging these on Goodreads as “read”. I don’t know if that’s kosher. The voice in my head has opinions. But I’m learning things I wouldn’t have learned otherwise, from books I wouldn’t have finished or started, with the same pleasure and actionability.

Maybe that’s cheating. But maybe the rules of our game have changed, and I haven’t learned them yet.

Update: 31 Dec 2025. Here are the books I read this year via AI summaries, along with links to the Claude summaries.

Sex at Dawn by Christopher Ryan, Cacilda Jethá Claude
The Molecule of More by Daniel Z. Lieberman and Dopamine Nation by Anna Lembke Claude
Noise by Daniel Kahnemann Claude
How to Change by Katy Milkman Claude
Lifespan by David Sinclair and The Telomere EFfect by Elizabeth Blackburn and Outlive by Peter Attia. Claude
How Minds Change by David McRaney Claude
The Obesity Code by Jason Fung Claude
The Diabetes Code by Jason Fung Claude
The Cancer Code by Jason Fung Claude
The Ants by Bert Hölldobler, Edward O. Wilson Claude
Price of the Modi Years by Aakar Patel Claude
Behave by Robert Sapolsky Claude
The Choice by Edith Eva Eger Claude
How Not To Die by Michael Greger Claude
Why We Sleep by Matthew Walker Claude
The Slight Edge by Jeff Olson Claude
The Origin of Consciousness in the Breakdown of the Bicameral Mind by Julian Jaynes Claude
Immune: A Journey Into the Mysterious System That Keeps You Alive by Philipp Dettmer Claude Gemini
The Gene: An Intimate History by Siddhartha Mukherjee Claude
The Structure of Evolutionary Theory by Stephen Jay Gould Claude
The Extended Phenotype by Richard Dawkins Claude
The Body Keeps the Score by Bessel van der Kolk Claude
The Culture Map by Erin Meyer Claude
Outcomes over Output by Josh Seiden Claude
The Song of the Cell: An Exploration of Medicine and the New Human by Siddhartha Mukherjee Claude
The Emperor of All Maladies: A Biography of Cancer by Siddhartha Mukherjee Claude
Reality is Not What it Seems: The Journey to Quantum Gravity by Carlo Rovelli Claude
The Happiness Hypothesis by Jonathan Haidt Claude
Why We Die: The New Science of Aging and the Quest for Immortality by Venki Ramakrishnan Claude
Same as Ever: A Guide to What Never Changes by Morgan Housel Claude
Build: An Unorthodox Guide to Making Things Worth Making by Tony Fadell Claude
Metabolical: The Lure and the Lies of Processed Food, Nutrition, and Modern Medicine by Robert H. Lustig Claude
The Language Instinct: How the Mind Creates Language by Steven Pinker Claude
How the Mind Works by Steven Pinker Claude
The Blank Slate: The Modern Denial of Human Nature by Steven Pinker Claude
The Stuff of Thought: Language as a Window into Human Nature by Steven Pinker Claude
The Demon-Haunted World: Science as a Candle in the Dark by Carl Sagan Claude
Entangled Life: How Fungi Make Our Worlds, Change Our Minds & Shape Our Futures Claude
Phantoms in the Brain: Probing the Mysteries of the Human Mind by V.S. Ramachandran Claude
Small Wins Every Day: 100 Powerful Ways to Transform Your Life and Health by Luke Coutinho Claude
The Tell-Tale Brain: A Neuroscientist’s Quest for What Makes Us Human by V.S. Ramachandran Claude

Update: 11 Apr 2026. I started adding this follow-up prompt:

Based on what you know of me, what are the things I should learn from these books? What might I do differently? Why? How?

Based on past conversations, AI knows me. For example, here were Claude’s thoughts:

Anand wants personalized life lessons from these books, filtered through what I know about him. Let me think about what I know:

Chief Innovation Officer at Straive, AI strategy, enterprise consulting

Faculty at IIT Madras, teaching Tools in Data Science to large cohort

Based in Singapore, Indian origins

Blogs daily at s-anand.net about AI experiments

Interests: data visualization, behavioral science, cross-domain pattern recognition, design

IIM Bangalore alumnus

Regular speaker at conferences

Core intellectual commitments: judgment and taste as scarce resource; verifiable/auditable AI outputs; Jevons Paradox applied to knowledge work; production costs collapsing shifts bottleneck to curation

Thinks in systems, economic, and forecasting frameworks with superforecaster-style probabilistic reasoning

Identifies as “LLM Psychologist”

Teaching an AI workshop for senior IAS officers at LBSNAA

That leads to high quality ideas. For example, it told me:

… you’re surrounded by people optimizing for the marathon — the visible, bounded deliverable. But Pheidippides’ actual run — the slow, unglamorous, unsexy one that actually won the war — was the one nobody watched.

What’s the Spartathlon version of your work? Probably: the long, patient work of changing how a client organization thinks, not just what tools it uses.

… make sure you’re spending some deliberate effort on it, even when there’s no one watching and no milestone presentation scheduled.

That led me to review transcripts for AI adoption impact. Answer: workshops help AI adoption most. I’m keeping an AI experiments list handy for impromptu workshops.

That’s real impact. In less time than I’d have taken to read a book.

Sure, I didn’t get what the books had to offer. I didn’t struggle. But I did get personalized value.

Misconceptions spread by textbooks

Sun, 30 Apr 2006 12:00:00 +0000

Misconceptions spread by textbooks. This site lists the facts, contrary to what most textbooks say about them.

Facts:

The ocean is blue because water is a blue substance. Not because it reflects the blue sky.

The sky is blue because air is blue.

Clouds actually remain aloft because they are warm inside.

Snopes on TV

Tue, 22 Jan 2002 12:00:00 +0000

Snopes is soon to be on TV.

The bunk stops here

Wed, 07 Nov 2001 12:00:00 +0000

A portal of urban legends. The bunk stops here.

Some rumours are true

Fri, 21 Sep 2001 12:00:00 +0000

On second thoughts, some rumours are true.

Urban legends

Tue, 18 Sep 2001 12:00:00 +0000

No updates till the weekend. I’m in Bangalore. In the meantime, remember not to trust anything that’s forwarded to you. It’s likely to be an urban legend.

Inverness

Sun, 05 Nov 2000 12:00:00 +0000

The next stop was Inverness. I didn’t know Inverness had any history to it. It wasn’t till we got there that I learnt that the Inverness castle was where Macbeth (of Shakespearean fame) ruled from. In fact, it turns out that Macbeth was a really nice king. There was this barbarian who fought him, lost, and turned to the English for help – who of course were delighted, and they killed Macbeth. This barbarian stupidly signed a document saying that Scotland would pay tributes to England, and that’s been the source of all the trouble.

The other thing to see, of course, was the famous Loch Ness monstor, or Nessie as the locals like to call it. We made one mistake, though. We landed there on a Sunday. As we got off the bus, we learnt that no tours operate on Sundays. The tourist information center was closed. I wanted to buy batteries for my camera, and the shops were largely closed. The only practical thing that was open was MacDonald’s, so we picked up a meal.

Fortunately, a guy called Tony Harmsworth came along on his van. He conducts guided tours to the Loch Ness, so we hopped on with about 8 others. Tony was apparantly involved with the Loch Ness centre since its founding, and had in fact headed it. So he was very knowledgeable about the history of the monstor.

Loch Ness in itself is beautiful. It’s a huge lake, very calm, and apparantly very deep. On one end is the sea. On the other end is the Urquhart castle. It’s a stone castle that was destroyed by the Jacobites (who had the habit of destroying everything they saw, actually). I have a sneaking suspicion Tolkien borrowed quite a bit of inspiration from the Loch Ness and the Urquhart castle. It could well be the remnants of Isengard – or the fort of the Uruk-Hai orcs.

The monster itself, of course, is just a myth. The local folks always thought there was a large fish in the waters. In 1933, Mrs. McRoy saw something large – about 6 to 9 feet – that she thought was a whale. Journalists caught on, and blew it up to a monstor. A Mrs. and Mr. Spicer claimed to have seen a snake-like monstor walking past the road. (We drove past this spot.) A vet student told his mother that he broke his bike because he fell off in surprise when he saw a dinosaur-like creature. Since he was a vet student, journalists believed him. A famous reporter found hippo pugmarks. In the end, it turned out to be a hoax using his hippo foot-shaped ash tray. Then this reporter rigs a photograph that looks like a dinosaur peeping out of the lake. That’s a hoax too, using a toy submarine and a piece of cardboard. Many of the other ‘monstor’ photographs turned out to be fakes – in one case, a labrador fetching a piece of stick (see if you can spot the face of the dog). In the 1987, operation Deep Scan searched the whole lake, and found 3 suspiciously large living objects, but they were no where near large enough to be a monstor. Sure, there’s probably some big fish down there, though.

The rest of the evening was at a local pub with some loud music on. Then we hopped on to the bus, reaching London at 7AM on Monday.

In time to get back to class.