Ideas We Believe

We are in the teenage years of AI . The playful toddler of 2021 (when models were cute) and the promising childhood of 2023 ( that AI will eventually replace us all) are well past us. Today we have to take the difficult call of whether the replacement of workforce that was promised is to be strategically acted upon. More than the risk of adopting the tech, lies the fear of missing out on the bus, since the wrath of failing to modernise is well documented. In 2025 alone, around a million jobs were lost to AI (that's what the company's memo read). And there is some merit to it. Many companies have publicly committed their AI investment, and have to now show the result of the massive bills - the unanimous answer to which has been productivity increase . But, such is the story of economics that one lever can be pushed without comprising on another. Efficiency is simple maths, the output must grow for the similar input (read workforce - and tip your hat to Karl Marx ). The sinister ...

The journey of Microsoft as a company can be paraphrased into three leaders and their approach to the industry. When bill gates developed a proprietary OS, he knew he had hit the potluck. Over the next few decades, he ensured that the Microsoft ecosystem kept itself updated and remained a close source. Given the philanthropy that Bill is known for, many expected the then-richest man to let go of the closed head and let the CS community have a look at the masterpiece, but Bill's wrath against open source was second to none. He and his successor Steve Balmer would call the open-source OS like Linux cancer. However, at the turn of the millennia, they were hit by rivals like Google in terms of the biggest tech company. Unlike Apple, Bill hadn't been able to trap its customers into an ecosystem. Therefore the battle moved away from OS to browsers, where the once reigning internet explorer faced an upward struggle from two emergent, one which its earlier rival had open-sourced to, an...

The rise of Chat GPT gave way to an interesting question, could the machine learning model come to pace with some of the benchmarks of human intelligence today. What followed next was the GPT models facing many management studies and legal tests and doing fairly worse in most. Chat GPT 4 today has increased its likelihood of passing the bar exam from a mere 10% to a whopping 90%. For someone who has seen how these models work, such a jump from one generation to another is no new. But it, therefore, poses more extensive and practical cases for us to implore. Before these large language models took the helm, any legal practice involved the tiresome job of scrounging through numerous precedents and preparing a case for both the defendant and the plaintiff. This involved rigorous search in databases (thanks to the digital revolution) and coming up with critical analysis from the texts. With AI promising to replace this final step in the process, it removes the final bits of human intellige...