AI Dial Tone

The Route to Simple

Yann LeCun, the Turing award winner in 2019 and current Chief AI Scientist at Meta, had an interesting observation: He stated that LLMs are the off ramp on the highway to ultimate intelligence. While GenAI can do astonishing things, it is not at the level of intelligence where machines actually think. He recognizes the significant milestone of GenAI, but we have not arrived at the final destination.

He observes:

…a teenager who has never sat behind a steering wheel can learn to drive in about 20 hours, while the best autonomous driving systems today need millions or billions of pieces of labeled training data and millions of reinforcement learning trials in virtual environments. And even then, they fall short of human’s ability to drive a car reliably.

And while we may be astonished today by GenAI capability, Andy Kessler, the venture capitalist and author, notes that GenAI is “a feature, not a product, and like every other technology, it’s starting to fade into the background.” He has a point. Look at the ubiquity of AI embedded in products, from Siri to Excel to Adobe. He observes that AI will be so common place, that it will fade into the background, make the complex simple, and become a commonplace expectation. Much like dial tone.

It was a Boston telephone operator strike in 1919 that ushered in the era of dial tones to replace humans. A pleasing tone delivered simplicity, musically asking callers “number, please,” hiding all the complexity of relay switches, tubes and amplifiers. AT&T was a hot stock in the 1920s as phones proliferated. Same for companies deploying other advanced technologies, such as radio. Eventually, investors stopped caring. Technology almost always vanishes from sight on its way to becoming productive and ubiquitous. Does that mean the end of the current AI hype cycle? Soon enough.

We think the insights from these two technology watchers should be heeded. Pace the investments in GenAI, recognizing it is still a horse race with more remarkable capabilities just around the corner. Keep the architecture modular, so that new components can be readily adapted where needed. Stick with Industry standards, and preserve interoperability. Make the complex simple, by leveraging the features of LLMs.

We work intensively with LLMs and business workflows for clients. Give us a call to discuss your best strategy for designing an AI architecture for your business. And remember that even dial tone, got 'dial toned' when cell phones took the market.