Posted by Aible ● Dec 2, 2019 9:44:15 AM

Future-Proof Your AI Investment


Anyone who tells you they know what AI will look like in five years is either lying or doesn’t know much about AI.

The truth is, it’s impossible to know what the “right” AI will be in the not-too-distant future. The best AI technology is evolving faster than even PC chips, which have developed at the blistering pace of Moore’s Law, with the number of transistors on a microchip doubling every two years.

At Aible, we don’t claim to know exactly what the future of AI will look like. The technology is moving too fast, and future market forces on AI are impossible to predict. It wasn’t long ago that SparkML was the hot new ML library. Then came TensorFlow. Next week, something else may well be the hot new technology.

That’s why Aible is serious about future-proofing your AI investment.

Aible is specifically designed to adapt to tomorrow’s AI technologies. You interact with Aible by answering straightforward business questions and then it automatically creates custom code specific to the underlying modeling technology. When the underlying technology changes, you can just refresh existing models to gain the benefits of future developments in ML with a single click.

This ability to adapt to new ML technologies is becoming increasingly important. Once companies deploy many predictive models, they face a critical problem each time a new ML technology is introduced: What do we do about the models we’ve already deployed that were trained on older technology? Asking your data science team to redo every ML project they have done in the last three years isn’t going to work. Your team could not have documented or remembered everything they did on the project, and some members may have left the company.

Aible, however remembers everything that was done, from the raw data to the deployed model. Aible never changes the raw data itself, all data manipulations and enhancements are done using auto-generated code. Aible stores users’ answers about their cost-benefit tradeoffs and business constraints that are used to automatically craft the custom loss functions. It remembers every decision the user made, such as excluding specific models or constraining the hyperparameter settings. Thus, when a new technology becomes available, all you have to do is retrain every deployed model with a single click. One click and you’re ready for the future.

It’s impossible to predict what the next new hot AI model, technology or infrastructure will be. With Aible, you don’t have to.