The notion of AI white box indicates AI systems with algorithmic transparency and human comprehensibility. Whereas AI glass box and Transparent AI can be more accurate and understandable nicknames, the concept is what is most important. This approach to AI and machine learning assures others that the algorithms act without bias and can be trusted.
More opaque ML algorithms (also called Black Box approach) can produce useful results that organizations and even governments adopt. Yet they do not guarantee that humans understand why they produced these results or decisions.
Black box AI is doing well thanks to the successes of Stable Diffusion and GPT-3 as well as programs based on them, such as Dall-E. There are also many of these tools available in the market that companies can use to develop their own applications. Because they’re built on neural networks or other opaque architectures, the inner workings of these systems—the processes by which an output is generated based on an input—are not explainable. The full potential of transparent AI is still unfolding and has not been fully unlocked.
As AI is introduced into more and more contexts with direct and significant impacts on humans – and as decisions become more obvious and visible, life or death or life changes – the opportunity for explainable, responsible and understandable AI increases. When humans see why an algorithm is acting the way it does, they can take the following actions:
- They can explain to the people affected by a decision how this decision was made, Explainable AI and Responsible AI two new buzzwords in the ever-changing AI lexicon.
- They can verify that the AI is not making decisions based on incorrect data.
- They can ensure that the AI does not ignore relevant information it has.
- They can find accidental and unwanted biases in the algorithms.
- They can identify biases maliciously introduced into algorithms.
AI Whitebox Use Cases
Any AI use case can theoretically be a white box AI use case. There is no reason for an AI system to have to be opaque or inexplicable. However, interest in transparent and explainable AI is highest in environments closely related to human well-being. These include the following uses of AI in decision-making and direct control of the physical world:
- make decisions at government level (eg whether to fund a new stadium);
- make law enforcement and criminal justice decisions;
- making medical decisions (eg, who should be allowed to take an experimental drug);
- make planning decisions;
- make important financial decisions;
- control moving vehicles (especially self-driving cars); and
- control of medical devices.
In these and many other situations, employees of government, banking, medical, police, and court organizations want to answer questions about how and why decisions were made and defend them as reasonable decisions. People living with the consequences of these decisions want understandable answers when they ask.
The future of white-box AI
Because there are fewer commercially available white-box AI systems to use to create a solution, organizations that need them must create their own for the foreseeable future. However, many university programs that teach AI, including Johns Hopkins and Michigan State programs, include courses covering AI ethics and courses teaching white box techniques. Some schools have institutes or research programs built specifically around white box principles, such as the Data Science Institute at Columbia or the Institute for Human-Centered AI at Stanford.
The extent of the future role of white-box AI depends on two factors. First of all, it depends on the accumulation of expertise and the evolution of the underlying platforms. As we’ve seen recently in the black box space, once there are enough qualified people, mature projects, and targeted startups in a space, progress accelerates sharply. Second, it depends on the evolution of laws and regulations governing AI in commercial and government spaces.
The implications of existing legal frameworks, like the GDPR, for how AI can and should work are still being worked out. New laws and agency rules will follow in jurisdictions around the world. If they tend to demand transparent, explainable and accountable AI, the white box approach will dominate the future landscape.
#Unlocking #Potential #WhiteBox #Machine #Learning #Algorithms #TechTarget