[ad_1]
Algorithms are increasingly the basis of our growth and our future. It picks up where human intelligence falls short in capacity, repeatability and predictive accuracy. The capacity to collect, analyze and categorize huge amounts of data. Repeatability in repeating this and learning how to repeat it to consistently deliver more accurate results. Predictability of previously unpredictable factors by using tons of historical data to predict the future and improve forecasting accuracy. Therefore, by applying these principles to a wide range of contexts and scaling exponentially, algorithms combat the transient nature of human engagement where results can vary from person to person depending on their abilities, biases, backgrounds and more. These traits usually give us the impression that they are unbiased compared to humans.
We now have ample evidence that AI/ML-based or algorithm-based models are not free of bias. The MIT Media Labs facial recognition software case is a well-known example. It didn’t detect darker-skinned faces primarily because the face samples fed into the system were mostly Caucasian male samples and that’s how the system learned. Google researchers found that even though China and India make up a third of the world’s population, they only account for 3% of the images in widely used datasets, while the US, with 4% of the population, accounts for 45% of the images!
A self-driving car crashing into an ambulance because the algorithm wasn’t trained on enough examples of how to slow down around an ambulance versus normal vehicles is another interesting example of bias.
Even in the financial services industry, online credit and digital insurance are very popular topics, but many studies point to an inherent gender bias in algorithms that leads to higher interest rates for women. Or the now “infamous” Amazon case where the AI ​​recruiting system eliminated most female job applications because historical data mostly shows successful men in these positions.
From school admissions, job acceptance to our credit rating to financial growth to our healthcare and even security, we are now deeply digitally embedded!
Despite the bias, algorithms are some of the greatest milestones of human achievement. But it would be simplistic and unwise to resort to rampant scaling and blind reliance on decision-making. Algorithms can complement human intelligence with their precision and ability, but it is not a blend of knowledge, emotions, experiences and ethics like the human mind. Making meaningful and fair decisions tailored to our individual needs and context is what differentiates our brain and makes it superior to even the most advanced algorithms
As the world prepares to be driven by artificial intelligence, digital India also needs to move forward smartly and lead cautiously, focusing on creating the right infrastructure and capabilities:
- Policy and regulatory frameworks to guide, guard against, and penalize illegal use: Especially in significant government-based applications that affect large populations or critical segments such as healthcare and security where even small mistakes can cause irreversible damage, we need good governance systems for algorithm-based decision-making.
- A check and balance mechanism with appropriate representative data to consistently mitigate bias: Feedback loops to correct skewed outcomes and avoid western-made pitfalls by using datasets representative of Indian traits and nuances. For example, financial algorithms in the west have inadvertently targeted certain communities in assessing their likelihood of loan default. We can mitigate similar unwarranted biases in discriminatory caste or community inferences.
- Responsible and compassionate human-driven decision-making that replaces algorithm-based conclusions alone. Strategic algorithms still can’t beat informed human decision making where empathy and emotion are key factors influencing decisions. This is strongly conveyed in the movie “iRobot” where the Robot decides to save Will Smith instead of the child from the car accident because he reckons he has a better chance of survival. It is of course a logical solution, but a human being with a moral and emotional perspective would not make that cold choice.
We need to train our workforce on how to use artificial intelligence properly, to be aware of its risks and potential impact, emphasizing the ethics of using it fairly and as a tool for making sound decisions. This is where the academic community could contribute by designing AI management training courses.
Even so, potential black swan events can occur that even the most sophisticated algorithms and management frameworks cannot predict. True intelligence is to be aware that we can never be aware of and correct all biases and mistakes, but to consistently seek and learn from them.
Diversity, complexity, digital acumen and innovation are India’s hallmarks. They are also what makes us a new age digital leader transforming major sectors such as government services, healthcare, financial institutions, education, infrastructure and more. Given the demographic difference, we have the opportunity to play a greater role in setting the benchmark for the wise use of algorithms for effective decision-making. This would give us the power to transform lives and protect personal freedoms by setting the stage for an inclusive and just digital age.
The author is a doctor of medicine at JM Financial.
Disclaimer: The views expressed are solely those of the authors and ETCIO.com does not necessarily subscribe to them. ETCIO.com shall not be liable for any damages caused to any person/organization directly or indirectly.
[ad_2]
Source link
