Many are waiting for the moment when someone announces the creation of AGI (Artificial General Intelligence), assuming two critical things:
- there is a universally accepted benchmark for what AGI truly is, and
- the majority of productivity and capability gains will come post-AGI, not from the advancements leading up to it.
The reality is that hundreds of billions (if not trillions) will be made on the journey toward AGI, whether we fully achieve it or not.
Take Amazon, for instance. They’ve been developing AI agents known as Amazon Q Developers, specifically designed to tackle software engineering problems.
SME-bench, a benchmark testing AI's ability to solve real-world software engineering challenges, shows that Amazon Q Developers can solve only around 38% of these problems. Many might look at this and think, "I knew it, AI is overhyped; I should stick to my traditional methods."
But with a well-defined problem and targeted training, Amazon Q Developers are already saving the company hundreds of millions of dollars:
"Amazon has migrated tens of thousands of production applications from Java 8 or 11 to Java 17 with the help of Amazon Q Developer. This has saved over 4,500 years of development work for more than a thousand developers (compared to manual upgrades) and resulted in performance improvements worth $260 million in annual cost savings."
Everyone envisions a Skynet-type AI system as the real disruptor. However, few recognize the baby steps that are already having a massive, quantifiable impact.
We can’t afford to wait and let disruption happen to us.
The AI transformation journey needs to start today, even with small incremental improvements, because by the time the "perfect" AI arrives, it will already be too late.
"Once a new technology rolls over you, if you're not part of the steamroller, you're part of the road."