Definition
Responsible AI is an approach to designing, developing, deploying, and using AI systems based on a frame work of principles that covers the broader societal impact of AI systems and the measures needed to reduce risks and negative outcomes that may arise with the use of AI, while still maximizing positive outcomes [1].
For further understanding
of this concept, let’s take a look at a scenario where a transportation ride-sharing
safety AI is given the challenge of ensuring passenger and driver safety in our
growing ride-sharing market while addressing gender-based concerns.
Using Responsible
AI approach, the following will have to be considered;
Safety First:
Real-time monitoring of unusual route deviations, detecting unsafe driving
patterns using local road conditions and provision of emergency alerts in local
languages.
Gender
Sensitivity: Optional female driver matching for women passengers
Cultural
Respect: Understanding local transportation customs and preferences and
cultural factors like avoiding certain areas during religious observances
Economic
Fairness: Prevention of surge pricing exploitation during emergencies and ensuring
fair pricing that accounts for local economic conditions
The basic principles of Responsible AI [2].
Origin
The origin of
Responsible AI can be trace back to 2016 when Amazon, Google, IBM, Facebook
(now Meta) and Microsoft formed an alliance on AI, with the sole aim to study
and promote the responsible use of artificial intelligence. This partnership set
the stage for today’s set of principles guiding responsible AI.
Since then, numerous public and private endeavors have joined the quest to refine voluntary guidelines for responsible AI. Some national and local governments such as US states (e.g., California, Colorado, Illinois, New York) already have laws in place related to AI, with more states set to follow [3].
Context and Usage
We can see the
applications and benefits of responsible AI in healthcare, businesses, customers
services, chatbots and human resources (recruitment).
In Healthcare,
organizations use responsible AI to improve patient outcomes while upholding
ethical considerations. Also, it helps treatment recommendation systems guide
treatment plans, eliminating discriminatory practices.
In Business and
Customer Service, responsible AI improves user experiences while maintaining
ethical business and customer service standards. Also, companies utilize
AI-powered chatbots to interact with customers without bias, ensuring
consistent and fair responses.
In human resources department, responsible AI is utilized to improve ethical recruitment and employee management. AI systems screen candidates impartially, ensuring fair hiring practices while hiring [4].
Why it Matters
With the
increase in popularity and use of software programs with AI features, the three
laws of robotics created by Isaac Asimov (a science fiction writer), is no
longer enough to be used as benchmark. There is need for standards in AI beyond
them. Responsible AI can be used to mitigate bias, build more transparent AI
systems, and increase user trust in those systems.
Usually, the data sets used to train machine learning (ML) models used in AI introduce bias via incomplete or faulty data, or from the biases of those training the ML model. When an AI program is biased, it can end up negatively affecting or hurting humans. For instance, it can unfairly decline applications for financial loans or, in healthcare, inaccurately diagnose a patient [5].
In Practice
Ada Health is a good example of a real-life case study of responsible AI in practice. Ada Health is an AI-powered chatbot that helps you assess a disease. The platform boasts of over 13 million users globally and over 34 million symptom assessments. The company practices responsible AI by securing data, implementing ethical principles, performing thorough tests, and making regular audits [2].
References
- Oleksandra. (2024). A Guide to Responsible AI: Best Practices and Examples.
- Vartak, M. (2023). Responsible AI Explained
- Convin. (2024). Examples of Responsible AI in Action Across Industries
- Hashemi-Pour, C., Gillis, A., S. (2024). What is responsible AI?