Scientists/Researchers Need More “Thinking Time”! Is Instant Communication Tech a Boon or Bane? Will AI-Powered Research Tools Help Researchers?

Dear Friends,

On July 25th, 2024, Nature published an interesting article titled "Scientists Need More Time to Think." While emails and instant messaging tools are essential to research, they also serve as major distractions. The article references a book by computer scientist Cal Newport, Slow Productivity, which discusses how science is becoming less disruptive despite an increase in the number of papers and grants. The paper suggests that knowledge workers need to slow down and spend more time thinking! It poses an important question: “What is the impact of lost concentration time on science—not just on the structure and process of science, but also on the content and quality of research?”

Ideally, a progressive academic institution should wisely identify and categorize faculty into four roles: Impactful Teachers (for insightful teaching), Deep Researchers (for disruptive ideas), Able Administrators (for effective management of limited resources), and Smart Mentors (for personalized advice and support based on students' socio-physical and psychological needs). There is no hierarchy among these categories; all are equally important in the age of AI-powered education.

We have access to excellent YouTube content that can replace a mediocre teacher. Tools like Scite_, a powerful research tool designed to assist scholars and researchers with enhanced literature reviews, are becoming essential. Writing assistants like Jenny.ai, Grammarly, and Jasper are increasing the scientific productivity of researchers.

Breakthrough unicorn startup Sakana AI from Japan has created the first AI Scientist. This AI can generate an idea and develop it into a full paper at a cost of approximately $15 per paper. The AI Scientist can perform idea generation, literature search, experiment planning, experiment iterations, figure generation, manuscript writing, and reviewing to produce insightful papers. Can you imagine how Sakana AI Scientist intelligently wrote a paper? https://sakana.ai/.../adaptive_dual_scale_denoising.pdf. This 11-page scientific manuscript, written entirely by AI, is formatted like a standard machine learning conference submission, complete with visualizations and all standard sections. The AI's performance is on par with that of an early-stage ML researcher. Thus, a mediocre researcher is also replaceable. A year's work for a researcher can now be done in minutes, allowing the generation of hundreds of medium-quality papers in a week.

AI mentoring tools already offer precise matching, data-driven insights, 24/7 availability, and cost-effectiveness. Hence, a mediocre mentor is also replaceable.

In such a scenario, unless an individual self-assesses where they fit among these four roles (Teacher, Researcher, Administrator, and Mentor), masters that area, and performs beyond mediocrity, their survival in the AI-matured phase is questionable.

That being said, the whole world is in flux right now. Universities, researchers, funding agencies, students, teachers, and parents are all navigating a complex and ever-changing environment. It's hard to determine what will work and what won't for some time.

Whether being a "Jack of All Trades and Master of None" or a "Master of One" is better is currently tough to judge.

If the AI Scientist continues to mature and surpass human performance levels, these old sayings may change once again! Let's wait and see, but this domain is poised for significant change!

Source: https://www.nature.com/articles/d41586-024-02381-x.pdf

Note: All are my personal views

Ravi Saripalle

Is the Theory of Market Structure Setting the New Norm and Inviting a New Model? Will it Impact Teaching in B-Schools or Tech-Schools? ONDC leads to ONDE?

Dear Friends,

Last week, I attended an important and intriguing discussion with the ONDC (Open Network for Digital Commerce) team, which I believe will be a revolutionary change in how we buy, sell, and deliver products or services in India! This innovation will not only influence commerce but also compel educators to rethink and possibly change how they teach in B-Schools or Tech-Schools.

Let us first understand the evolution of market structures. In early civilizations, trade operated through a simple barter system without formal market structures. Transactions were local and based on the direct exchange of goods and services. Later, formal markets emerged in towns and cities, often characterized by guilds and monopolistic trade practices.

During the 16th to 18th centuries, monopolies (where a single entity controls the supply) and oligopolies (where a small number of firms dominate an industry) were prevalent, particularly in colonial trade. The 18th to 19th centuries, especially during the Industrial Revolution, saw the rise of competitive markets, though certain capital-intensive industries exhibited oligopolistic characteristics. By the late 20th century, globalization and technological advancements led to the formation of global oligopolies.

In the 21st century, the rise of digital platforms and network effects created new forms of monopoly and oligopoly, along with instances of perfect competition in some areas, often regulated by antitrust actions and government policies. However, the maturity of technology has led to a mix of highly concentrated digital markets and fragmented niche markets driven by new technologies and consumer demands.

In this context, India is introducing the concept of the Open Network for Digital Commerce (ONDC), which is comparable to UPI but in the realm of commerce. Today, if you want to book a flight, you use one app; if you're going to order food, you use another; if you want to buy a suitcase, you use yet another app. However, ONDC will allow you to do all this on a single platform, enabling you to compare prices across all sellers and choose the best delivery agent, all within a single app!

This platform isn’t just beneficial for buyers; it also provides sellers the opportunity to reach a global customer base with significantly reduced marketing or customer acquisition costs. In India, more than 12 million sellers earn their livelihood by selling or reselling products and services, but only 15,000 sellers (0.125%) have enabled e-commerce. With around 936.16 million internet subscribers in India, including 350 million mature online users, it is nearly impossible for a small roadside seller to tap into such a vast e-commerce market. However, ONDC offers them equal opportunity.

ONDC recognizes the unique potential to increase e-retail penetration from the existing 4.3% to its maximum potential in India. With 26 banks, BSE, and NSE investing in this platform and ONDC being established as a Section 8 not-for-profit company, the initiative is poised to make a significant impact.

ONDC’s architecture allows Buyer Apps and Seller Apps to manage order fulfillment as they see fit. This is good. Moreover, ONDC, as the orchestrator, has no direct role in—or even visibility into—any transaction. This is better. The introduction of scoring, badging, and a user-driven system to evaluate the quality of products and the performance of sellers and logistics service providers is the best approach to preventing online malpractice.

However, I have one lingering question. What if a large player takes advantage of a small seller, especially one in a remote part of the country or an early-stage startup doing local business? With the advancement in AI, this large player could gain global access to all sellers displayed on the platform, including their pricing, margins, delivery methods, customer psychology, and more.

Do you think teaching in the classroom is set to change with the introduction of this platform? It’s no wonder that when teaching becomes a commodity and an “Open Network for Digital Education (ONDE)” is introduced, every course, certificate, and student will operate similarly! That day might not be far off!

ONDC leads to ONDE?

Ravi Saripalle

Near Miss (Silver Medal with Low Spirits) vs. Close Call (Bronze Medal with Positive Spirits): A Psychological Syndrome Applicable in Education

Dear Friends,

Last week, a well-wisher forwarded me an interesting WhatsApp message about the psychology of a silver medalist versus a bronze medalist. Surprisingly, research shows that a bronze medalist is generally happier than a silver medalist at the end of the game. Ideally, one would expect the silver medalist to be happier than the bronze medalist, but the human mind doesn't operate like mathematics. This phenomenon is due to counterfactual thinking: the silver medalist often thinks, "Oh, I couldn't win the gold medal," while the bronze medalist thinks, "At least I got a medal!" The silver medal is won after losing (near miss), but the bronze medal is won after Winning (close call)—hence, the contrasting emotional responses.

Reflecting on this, I was reminded of my own experiences. I attended a government municipal high school where the competitive environment was not as intense as today. We were more relaxed and didn’t worry much about rankings. However, my parents were teachers at the same school, so I felt some competition. I ranked first in my 6th, 7th, and 10th grades, but in the 8th grade, I came in second. After reading the message about counterfactual thinking, I realized I had felt a small sting from securing second place back then. Yet, I wasn't too concerned in the 9th grade when I ranked fourth. That Freefall (slipping from 1^st to 2^nd to 4^th) of ranks led to a free fall of worry, and then that process helped me refocus and excel again in the 10th grade. Of course, after that, I didn’t chase or ever receive gold, silver, or bronze but instead maintained my own pace—call it steel, aluminum, copper, zinc, lead, or what have you!!!

I have spoken to many JEE rankers over the years and noticed a similar pattern of counterfactual thinking among them. Those in the 99.9+ percentile are generally happy because they’ve secured spots in their preferred Tier 1 institutions. However, students scoring in the 95-98 percentile are often less content because they missed out on their dream seats, despite still being in Tier 1 institutions. Interestingly, those in the 80-90 percentile range often feel better than those above, as they are happy with a decent college and branch, even if it's at a Tier 2 institution.

This dynamic applies to life situations as well. I’m sure we have all experienced something similar. Some of us may show it openly; others may accept it quietly, while others detach themselves from the situation. Research also suggests that collective groups receiving silver medals respond more positively than individuals in similar situations.

Though I’m not claiming to be an expert or mature enough, I believe this research can help us manage such situations more effectively.

Counterfactual Thinking: Why near misses leave us discontent while close calls fill us with relief—and how this psychological insight applies to education and life."

Note: These are my personal views

Ravi Saripalle

From the Track to the Streets: Climate Change Challenges Us All! Read this perspective!

Dear Friends,

 

I live in a tropical climate characterized by warm temperatures and high humidity throughout the year. Our nights from October to February are cooler, while the days from March to June can exceed 40°C. Thanks to climate change, we sometimes experience temperatures of 45°C and above! Most of the time, our work is conducted under a roof in a sophisticated environment.

 

Now, let’s review this scenario from the perspective of a player. According to Gardner's Multiple Intelligences theory, the player is born with kinesthetic intelligence, also known as bodily-kinesthetic intelligence, which refers to the ability to use one's body effectively and with a high degree of skill. Individuals with strong kinesthetic intelligence are often proficient in physical activities such as sports, dance, acting, or anything that requires fine motor coordination.

 

Let’s consider the situation of a top athlete (high kinesthetic skills) born in this region, who, after training intensively for 14 to 15 years—practicing 14 to 15 hours a day—finds themselves competing in a climate that is the exact opposite. Can you imagine the acclimatization (natural environment) issues they would face though they went through acclimation (artificial environment setup for training)? 

 

The body may struggle to regulate temperature, and moving from a low-altitude to a high-altitude environment (or vice versa) can affect breathing and stamina. Muscle fatigue, recovery time, and psychological stress are just a few of the many factors that could influence their performance.

 

This is the reality for every Olympics, especially for the Paris Olympics. Athletes at the 2024 Paris Olympics compete against the world's toughest opponents, the clock, the record books, and against increasingly extreme heat. You can imagine police officers cooling off under “misters” (which release water droplets into the hot air) at various locations around the venue! Consider the sequence of events this year: rain soaked the opening ceremony, followed by an extensive heat wave across France that reached 35°C, a temperature more typical of tropical regions!

 

Let’s reflect on the Tokyo Olympics held in 2020, which recorded the highest temperature of 32°C—considered the hottest Olympics in history. More than 110 athletes suffered from heat-related issues (Vox.com, Aug 9, 2024).

 

The next Olympics are set to be held in Los Angeles, USA. Today, the Olympics are not just a sporting event; they represent a livelihood for many, a part of sports tourism, and a business. So why not shift the Olympics to winter? Doing so could affect sponsorships, as summer holidays worldwide boost viewership.

 

Climate change must increasingly be seen as an existential threat to sport. It’s not just the Olympics at risk; many livelihoods are also on the line. For example, the impact of heat stress on informal sector women in India or other tropical countries is exceptionally high. These women work 7 to 9 hours daily under direct sunlight. You can imagine this segment's slow productivity and income loss.

 

Heatwaves Don’t Discriminate: From Gold Medals to Daily Wages! What do we have to raise to combat this discrimination?

 

Ravi Saripalle

While the world is still trying to understand “Machine Learning” (ML in AI), the industry has started “MUL- Machine Unlearning”! Why?

Dear Friends

I could not speak or pronounce certain words in my mother tongue during childhood. It used to be a herculean task for my parents and teachers to correct them. My Telugu teacher used to say, first time pronounce right, else it becomes an unerasable language tattoo! As it is language, we can still adjust and compromise as it does not hurt anyone or doesn’t do any big damage.

What if AI makes this mistake with false information, with a bias, with a hallucination? To date, AI is busy learning, learning, and learning.

Villalobos et. al's (2024) findings indicate that if current LLM development trends continue, between 2026 and 2032, LLMs could use nearly all available public human text data for training or slightly earlier if models are overtrained. Llama 3 is pre-trained on over 15 trillion tokens collected from publicly available sources. The total amount of English literature could be around 40 to 90 trillion tokens. If all languages are combined, it might reach 100-200 trillion tokens. The ratio of tokens per word is approx. 0.75. “ChatGPT needs to ‘drink’ a 500 ml bottle of water for a simple conversation of roughly 20-50 questions and answers. Can you understand the training and usage costs of each question we type?

Having said that, the total number of words heard by humans by the age of 20 would be approx. 100 – 200 million words (educatingsilicon.com

). The maximum learning would be minuscule compared to these LLMs. With this massive learning rate, if AI misunderstands a particular issue, Unlearning becomes costly!

What is Machine Unlearning? It refers to removing the influence of specific training data points on an already-trained machine-learning model. It eliminates data points for fairness and accuracy.

Let's make some assumptions. It estimated that the cost to train GPT-4 was about $100 million. The same training costs exist if we want to make specific contexts unlearn (confidential datasets). It is the exact energy requirement. It produces sizable carbon emissions. Hence training and untraining are both costly.

While discussing these algorithms, technologies, training costs, and Water footprint, nobody talks about Human Unlearning Algorithms for all our mistakes, biases, misconceptions, and so on! To some extent, Spiritual Teachings act as these unlearning algorithms in humans, but again, if an unauthorized teacher teaches, it becomes unwanted training, another unsolved circular puzzle to solve!

Still, the World never stops learning while unlearning!

Ravi Saripalle