Posted by Arijit Sengupta ● Apr 30, 2020 3:37:35 PM

In Turbulent Times, You Need AI That’s Flexible.


In Turbulent Times, You Don’t Need AI That Gives You a Single Perfect Answer. 

You Need AI That’s Flexible.

With businesses facing unprecedented uncertainty, some companies have turned to AI looking for “an answer.” But these businesses are already on the wrong track – ¬any AI that produces a single answer for how to navigate the months ahead is bound to be wrong. The fact is, business assumptions and expectations are changing so fast that any answer AI comes up with will soon be invalid. 

Businesses can’t confidently settle on a single strategy because traditional business tools are for now essentially useless.  Business intelligence tools used to be a great way to uncover hidden patterns in existing data in order to map out strategy. But BI is backward-looking, and in the current environment, the past looks nothing like the present, and will look even less like the future. Traditional AI relies on data alone, but everyone’s data is now woefully out of date. Quarterly planning is no help either, because it’s impossible to predict what will happen next month, much less next quarter. 

The answer is that there’s no single answer. And the way forward is all about flexibility. 

Businesses need to let go of the natural impulse to find an answer and focus instead on having the flexibility to change provisional answers very quickly as new information surfaces. The best way to navigate the current uncertainty is to have a series of flexible responses that adapt quickly to changing circumstances.

That’s especially true for AI. Traditional AI is built on the belief that advanced mathematics and data science can produce finely-tuned predictions that accurately forecast the future. But turbulent times demand a very different approach. 

What’s important now is how flexible the AI is and how quickly you can change the AI to give better answers given new information. How many levers of flexibility are there in the AI? How broad is that flexibility – does it go from the beginning of the AI process to the end and back again? The more friction points and manual steps there are in the process, the less flexible the AI is. The fewer inputs it considers, the more rigid it is. The more expertise that’s needed to translate the inputs into outputs, the less adaptive it is. 

Aible is built from the ground up to be flexible. Flexibility isn’t an add-on feature of Aible – it’s baked into the platform. Aible ensures maximum user flexibility in three important ways:

  1. Aible is fully automated while at the same time, taking in human input.
    Aible is automated end-to-end, from raw data to the predictive models deployed in the enterprise applications business people use every day, such as Salesforce and Tableau. At every step, Aible captures feedback at scale, creating a powerful feedback loop that stretches down to the end users and back up again to management. Aible can do everything automatically but humans have the ability to adjust the recommendations, based on the business realities they’re seeing on the ground. This is the time for augmented intelligence, where you use human domain expertise and assumptions about the future along with AI to bridge the data gap. With augmented intelligence, you can take advantage of automation while at the same time using human input to adjust quickly to changing conditions.
  2. Aible enables people to consider many possibilities.
    Aible lets users try out dozens or even hundreds of scenarios to see how different assumptions would impact the business. Users can go in and change assumptions, and multiple people can collaborate on different scenarios. Aible creates an efficient frontier of many models that would be optimal in each of those scenarios (plus thousands of other scenarios that haven’t been manually considered), deploys the models automatically, and seamlessly uses the right model given the specific circumstances at the time of prediction. At every step, the process is flexible and preserves options. Users aren’t locked into testing just a few scenarios or deploying a single model. That’s critically important at a time when the range of possible business scenarios is vast.
  3. Aible takes feedback at scale and responds quickly to change.
    Aible enables people on the front line to provide feedback based on the business realities they’re seeing. End users provide feedback when business assumptions are no longer correct or if the AI is telling them to do something that doesn’t align with their preferences or what they’re encountering day-to-day. That feedback in turn informs management so that micro-adjustments can be made collaboratively. In this way, Aible constantly learns from every prediction it makes.

Unlike traditional AI, this built-in flexibility makes Aible particularly well-suited to help businesses navigate uncertain times. Traditional AI requires a lot of manual consulting and expertise for every one of the above steps, and can’t take feedback at scale. Conventional AI is a top-down process in which a handful of experts create an AI model and the rest of the business passively consumes it, rather than continuously collaborating and adjusting. 

Businesses can’t afford AI that inflexibly produces a single answer rather than a portfolio of models that covers a wide range of possible business realities. In the current environment, today’s answer may be obsolete next week. Aible empowers businesses to adjust quickly to change – and the only thing businesses can count on for the foreseeable future is change.