💣The 2 biggest AI myths you must watch out for

Aug 06, 2021

Hello,

Today, it’s the 91st anniversary of Virakesari. This was the newspaper founded by my grandfather after he immigrated to Sri Lanka from India. On August 6th, 1930, the first edition of this Tamil daily was printed. He served as its founding editor and ran the operations for over 20 years. In the 1950s, he permanently moved back to India amidst political instability in Sri Lanka.

Today, Virakesari, or “Victorious Lion” as its Tamil translation goes, continues to be one of the largest circulated Tamil newspapers in Sri Lanka, with 120,000 daily copies. While we no longer have a connection to the newspaper, two things still influence me. My name - ‘Kesari’ became our family's surname after the newspaper was founded. Secondly, perhaps, it's no coincidence that I'm strongly drawn to writing.

This newsletter will take you about 4 minutes to read.

I. Spotlight: 💣The 2 biggest AI myths you must watch out for

A few months ago, I was on a Zoom call with the Head of Technology of an Asian retail conglomerate. His team had ownership of BI (Business Intelligence) reporting across the organization.

The team aspired to improve its data analytics maturity and had reached out for help.

The technology head was curious about data science. “How can we make the transition from BI to Artificial Intelligence (AI)?”, he asked me.

This is a common question I get from executives. While the intent is good, there are two key misconceptions here.

And, this is the root cause of a lot of wasted investments in AI. Let me explain.

(Photo by Michael Dziedzic on Unsplash)

Myth 1: AI can replace BI

AI is a powerful technology. But it can’t do everything.

Different decisions need different kinds of insights. Think of analytics as a toolbox that includes a collection of tools. Broadly, there are four kinds of techniques:

a. Business Intelligence: If you’re running a retail business, your teams must know, on a daily or hourly basis, which products were sold. BI reporting can satisfy this need with simple, descriptive insights. This historical view is crucial to run the business.

b. Statistics: For deeper answers from data, you will need stronger tools. Is the newly launched “product A” eating into sales of your bestseller, “product B”? Statistical techniques can help you perform diagnostic insights and conclude with confidence.

c. Simple Machine Learning: Let’s say you want to know where to spend your marketing budget. Simple machine learning algorithms can help you. Predictive insights from market-mix modeling can show how marketing initiatives impact sales.

d. Artificial Intelligence: To find how effective is your store signage in attracting footfalls, you can use AI. A crowd-counting algorithm can estimate your footfall. Combine this input with sales numbers from BI and you get descriptive insights on conversion. AI can help with diagnostic and predictive insights as well.

So, it’s clear that you need many tools in your toolkit. Depending on the business problem at hand and the data available, you must pick the best tool for the job.

Myth 2: AI is a sign of higher analytics maturity

There’s one challenge with the pyramid visual we’ve seen earlier. People mistake AI’s higher position to be a sign of better maturity. If a company is doing projects on AI, has it leveled up?

Not really. I’ll share two reasons.

Firstly, a company’s ability in predictive analytics is no guarantee that it would excel at descriptive reporting. I’ve come across organizations that have not ventured beyond BI, but their reporting is far more robust than those with AI models in production.

Secondly, techniques are not the sole determinant of your analytics maturity. There are more important factors such as soundness of data strategy, alignment with org vision, quality of data, the strength of processes, ability to productionize, and user adoption.

It is this combination of factors that characterize the data and analytics maturity of an organization.

It’s just incidental that organizations acquire AI capabilities later on in their journey. AI is not just complex in the early stages, but it is relatively less useful than BI for the needs of that phase.

So, when people ask me about replacing BI with AI, I remind them, “Once you learn to run, you don't stop walking.”

II. Industry Roundup:

1. How are startups using AI to tackle climate change?

27 minutes | Forbes | Rob Toews

Deep learning is often blamed for guzzling compute power leading to huge carbon emissions. However, many AI startups have sprung up to help fight climate change. This article lists the novel approaches adopted by them spanning 8 critical areas such as carbon accounting, precision agriculture, climate insurance, and prediction of fires. An excellent article if the intersection of AI and climate change interests you.

-> Read the Article

2. Putting the power of drug discovery into the world’s hands

2 minutes | DeepMind | Demis Hassabis

Deepmind’s AI system, AlphaFold, solved the 50-year protein folding challenge last year. This opened up possibilities of big breakthroughs in drug discovery. What’s more impressive than this historic milestone? In recent weeks, DeepMind open-sourced AlphaFold’s code and made public the database containing 3D structures of almost all proteins in the human body. Amazing work and great gesture, isn’t it?

-> Check out the tweet thread

III. From my Desk:

1. Article: How analytics helped this hospital beat the Covid-19 upsurge

7 minutes | Forbes

The pandemic crippled most healthcare organizations. This case study shows how data & analytics turned mission-critical in Parkland Hospital's fight against Covid-19. The article traces the organization’s path of analytics maturity and shows how the leadership systematically built agility and resilience.

-> Read the Article

2. E-book: How to drive data-led transformation using data maturity

7 minutes | Gramener

Where should you begin on your organizational transformation with data? How can you assess if you’re moving in the right direction? Data maturity can be an invaluable tool in this journey, quite like a compass to an explorer. This ebook explains the concept with examples.

-> Read the E-book

Tired of all the jargon in the data world? Let these puppets bust your stress 😆

Thank you for subscribing and reading the newsletter. I appreciate your attention,

Ganes.

PS: Did someone forward this to you? You can subscribe here.

My Website | LinkedIn | Twitter | YouTube

I’m Ganes Kesari. I publish ‘Data-Driven Future’ to help understand how data shapes our world, explore key trends, and explain what they mean for you today. I speak and write to demystify data science for decision-makers and organizations.

Recent Issues:

Our Data-driven Future

Discussion about this post