Math Vs. Magic

In May 2024, while releasing one of the latest updates to his company's flagship model, Sam Altman tweeted "Feels like magic to me."

If you open your browser screen and go to chat.openai.com , just look at the bottom of the screen and regardless of which version of the model you choose to use, whether its a premium plus subscription or the $200K a day bleeding freemium version, you will find a footer caption saying "ChatGPT can make mistakes. Check important info."

So does Mr. Altman wants to say is the magic his company built is not reliable enough? Even after paying him $20 a month for a plus subscription, the caption doesn't go away. So he is going to the next investment round with billions of dollars in the ask, and a minimum ticket size of $250 million with a disclaimer saying -"ChatGPT can make mistakes. Check important info."

Or maybe Sam was talking about the quote from science fiction writer Arthur C. Clarke, about magic of sufficiently advanced technology.

"Any sufficiently advanced technology is indistinguishable from magic." -Arthur C. Clarke

Magic or Math debate

The current excitement and enhanced interest in artificial intelligence is driven by the numerous large language models which are in a race to tip another one off the charts. This race was kicked off by the first GPT launch by OpenAI in November, 2022.

But is it actually magic or just too much math, which will need some skill development and practice to comprehend?

So there are two dichotomous viewpoints proposed by the same person. If Sam Altman wants to raise a large round with hundreds of billions of dollars in capitalisation as the draw, I guess the investors will need to understand the math behind the model, in order to perform due diligence on the prospective investment. In this scenario, just like their other investments, the investors are well within their rights to discredit the magic potion and ask some scientific questions on the product front.

"We certainly haven't solved interpretability."- Sam Altman

The statement of Arthur C. Clarke wasn't focused on magic, it was about sufficiently advanced technology. But as it happens, in a world mired with stressors and challenges, muzzled by the complexity of damaged personal relationships, workplace toxicity and government taxes- people seek respite by focusing on magic, which excites us and gives us hope that tomorrow will be somehow better.

The math vs. magic debate is a crucial one in the current times, in order to allay the fears of a superhuman intelligence taking over, and in order to ensure responsible AI.

What's wrong with Magic?

You can't trace the genesis of magic, simple. You cannot control magic, as you do not understand how magic has happened. The statement like above from Sam Altman, are largely responsible for the fear and anxiety, the business community feels in embracing AI as a strategic enabler for their growth.

There is no doubt that AI is going to be an extremely crucial technology from a future standpoint, and all companies/executives must think about how it can solve some of the most pressing problems for their business. But we all know- there is no magic in business, its all math and too much of math.

We can control math, since we know its component parts, the equations which form math-but we cannot control magic.

"Feels like magic" is a great marketing catchphrase, you can actually sell anything from tissues to chocolates to a private jet, using the same phrase.

But with the disclaimers like the ones OpenAI has placed along with 'magic', it sounds like they are asking, "Give us your money, as a customer, investor since we have created something that was never created before, but we cannot take any responsibility for its effect, should you choose to use it in your day to day work."

How to proceed then?

Business executives responsible for technology decisions must question every sales rep or consultant or self-proclaimed AI guru, to explain the math behind the magic before signing on the dotted line. Even real magicians have tricks, laws of physics that they use in order to create an illusion.

When we talk about AI we are talking about a really relevant and intelligent technology, which can do a lot of things which a human can do- so it is a mimic of human intelligence, in a bounded rationality environment. If the creators of a certain technology solution do not know how it works, where will a customer of such technology go, if the same solution does some damage to the business, customers, or investments.

Without understanding the basis on which an AI based investment advisor is making recommendations to me for the next investment I have to make in the stock market- can I procure the solution? Without knowing the basis on which an AI based Health advisor makes recommendations for my next meal, can I go ahead and put my health in that AI's hands?

Fear of unknown is the most primordial fears we have carried for generations, total and complete knowledge can offer an effective remedy. It is extremely important to focus on maturing the emerging technology enough, so clear answers are available and explainability can be established. For the products, leveraging some base level AI like prediction models, with clear statistical explainability- a timely investment into them is wise and worth a try to reduce costs, accelerate revenue and foster growth.

But for the magicians out in the dark, the message is to refine their craft and make it more transparent and accountable. The business value of magic is for the magician, the seller- for others, what matters more is the pressing problems they are facing day to day.

-Mohit Sharma

This piece is written without prejudice towards any individual or company. Any sources referenced have been directly attributed and are owned by the respective third-parties. The insights I share are based on my own personal experiences on the journey I have been fortunate to live. -Mohit Sharma