Topics In Demand
Notification
New

No notification found.

White Box AI vs Black Box AI
White Box AI vs Black Box AI

May 17, 2021

AI

106

0

As we increasingly rely on artificial intelligence systems to support human decision making, businesses are under pressure to produce interpretable AI.
 

What Is Black Box AI?

Black box AI is where AI produces insights based on a data set, but the end-user doesn’t know how. Machine learning programs reach conclusions from the data input, but it’s not clear how the program came to them. These approaches used to be the industry standard for machine learning, but things have changed.

“Opacity is the heart of the Black Box Problem—a problem with significant practical, legal, and theoretical consequences. Practically, end-users are less likely to trust and cede control to machines whose workings they do not understand.” - Burrell, 2016; Ribeiro, Singh, & Guestrin, 2016

Given this mystery, it’s easy to understand why many companies have moved away from black box AI. But it has its advantages. The outputs of black box AI tend to be remarkably accurate. This accuracy comes from the algorithms’ complexity, but this also results in their lack of transparency.

Deep learning algorithms often employ a black box approach. These neural networks can be so complex that humans can’t explain the outcomes, even if they prove accurate. They can produce some of the most groundbreaking results of any AI type, but even their developers don’t understand how.
 

What Is White Box AI?

In contrast, white box AI is transparent about how it comes to its conclusions. A data scientist can look at an algorithm and understand how it behaves and which factors influence its decision-making. As people have grown increasingly suspicious of black box AI, these models have risen in popularity.

White-box models (WBMs) provide clear explanations of how they behave, how they produce predictions, and what variables influenced the model. WBMs are preferred in many enterprise use cases because of their transparent ‘inner-working’ modeling process and easily interpretable behavior. For example, linear models and decision/regression tree models are fairly transparent, one can easily explain how these models generate predictions. WBMs render not only prediction results but also influencing variables, delivering greater impact to a wider range of participants in enterprise AI projects

White box AI tends to be more practical for businesses. Since a company can understand how these programs came to their predictions, it’s easier to act on them. Businesses can use them to find tangible ways to improve their workflows and know what happened if something goes wrong.

Since white box AI insights tend to be more linear, they’re often less radical or disruptive. They can still lead to reliable and helpful predictions but usually won’t provide out-of-the-box, game-changing ideas. While they may not be as technically impressive, their transparency does provide a higher level of reliability and trust for the end user.

WBMs and Impact on User Persona

There are three key personas to consider when applying ML to solve business problems: model developers, model consumers and the business unit or organization sponsoring ML initiative. Each persona has a different priority and implications based on the specific modeling approach.  Model developers care about explainability, model consumers care about actionable insights and for companies and organizations, the most important attribute is accountability
 

Can Black Box and White Box AI Work Together?

Both white box and black box AI have unique strengths and differences. As a result, you can’t necessarily say one is better than the other in every circumstance. Many machine learning endeavors today try to balance the two to achieve both interpretability and accuracy.

As AI now plays a prominent role in society, trust and transparency become increasingly crucial. Consequently, many companies have favoured white box approaches in recent years but haven’t abandoned black box AI entirely. While black box AI may be unsuitable for highly regulated industries, it is still incredibly useful for other AI models.

Machine learning models require rigorous testing, but a program’s developers may not be able to test it without inflicting bias. Instead, developers can turn to third-party black box testing as a quality control solution.
 

AI Is a Varied and Continually Changing Field

AI encompasses a vast range of different techniques and technologies. It is also continually growing, which makes it hard to keep up with. But this growth is for the better. In instances like white box vs. black box AI, different approaches have produced a sum better than its parts.

In understanding the difference between varying AI applications, we can see where technology is going next. When we know where AI methods come from, we can understand where they’re going.


That the contents of third-party articles/blogs published here on the website, and the interpretation of all information in the article/blogs such as data, maps, numbers, opinions etc. displayed in the article/blogs and views or the opinions expressed within the content are solely of the author's; and do not reflect the opinions and beliefs of NASSCOM or its affiliates in any manner. NASSCOM does not take any liability w.r.t. content in any manner and will not be liable in any manner whatsoever for any kind of liability arising out of any act, error or omission. The contents of third-party article/blogs published, are provided solely as convenience; and the presence of these articles/blogs should not, under any circumstances, be considered as an endorsement of the contents by NASSCOM in any manner; and if you chose to access these articles/blogs , you do so at your own risk.


images
Nishant Kumar
Technology Enthusiast

© Copyright nasscom. All Rights Reserved.