Don’t let deep-fake generation act like an AI exemplar
The AI industry must showcase deployments of AI that are clearly doing people at large plenty of good
Looks like Geoffrey Hinton’s words of dissonance about the Artificial Intelligence (AI) industry have opened the flood gates. Over the past fortnight, we came across more negative write-ups about the industry than ever before. The important point is that these criticisms are coming in from significant people. An article in The Atlantic, ‘AI is About to Make Social Media (Much) More Toxic: We must prepare now,’ was by Jonathan Haidt and Eric Schmidt. The former is the Thomas Cooley Professor of Ethical Leadership at the New York University Stern School of Business and author of several famous books. Schmidt is the former CEO and chairman of Google. The article ‘AI Is a Waste of Time’ in The Atlantic was written by distinguished journalist Derek Thompson. ‘Will AI become the new McKinsey?’ in The New Yorker was by Ted Chiang, yet another distinguished journalist. Doug Rushkoff, who in his interview last week to Wired magazine renounced the digital revolution, was one of the original enthusiasts of digital technology’s pro-social potential. He introduced words like ‘digital natives’ and ‘viral media’ to the vocabulary of the technology world.
Not only are many significant people stepping forth to talk about the potential dangers of the AI industry, there is also a change in the tone of the criticism. It is becoming harsher. The points raised by Haidt and Schmidt in their article that the AI technology can be misused by vested interests to create more misinformation and disharmony in societies is definitely a serious issue. But the New Yorker article raises even more profound questions. It asks whether AI technology is a tool that those with money will use to make more profits off the effort of others. Will it encourage an ever-growing concentration of wealth among an even-smaller number of people? Further, can the AI industry ameliorate the inequities of our world and raise the living standards of the common man?
In the midst of these growing criticisms of the AI industry, what surprises me most is the broad silence of its apostles. There is no one of prominence who is stepping forward to provide responses to the various criticisms being levelled against the industry. This silence does not bode well for it.
The news that AI tools can create an image of Pope Francis wearing a white puffer jacket has made headlines around the world. But does this technology that makes fake images do anything to improve the lives of the common man? Instead, such AI tools will only reinforce what Derek Thomson said in his article: that many people today are simply using AI tools merely as toys. Such AI technologies reinforce the growing belief that AI technology will mostly be used for making deep fakes and pushing falsehoods, and will thus create disharmony around the world. So AI tools of this type could end up doing more harm to the industry than good. Unfortunately, it is exactly these generative AI technologies that are hogging the news headlines everywhere. The AI industry therefore needs to project use cases that have made a genuine difference to the lives of common people.
Tuberculosis (TB) is one of the most infectious diseases in the world. TB claimed 1.5 million lives in 2020 alone. Prevention and early diagnosis of tuberculosis are key to its treatment. Chest X-rays are the most sensitive and cost-efficient way to screen for tuberculosis and other lung diseases. However, there are not enough trained radiologists to interpret every chest X-ray on time, leading to delays in TB diagnoses. AI companies like Qure.ai have developed technology products that can read chest X-rays far more accurately than what radiologists can do. This use case is not about AI making radiologists redundant. But it is about providing a radiologist’s services in very many places around the world where radiologists are not available.
Use cases of AI technology making a real difference to millions of people, even those at the bottom of the pyramid, can be found in India’s Digital Public Infrastructure (DPI) initiatives. Our DPI schemes span sectors like financial inclusion, healthcare, education, agriculture and many other public services. What is interesting is that Indian DPI schemes also deploy sophisticated AI technologies, such as machine learning, conversational AI tools and image analytics.
For a long time, Among the biggest challenges of biology was solving the protein-folding problem of understanding the molecular structure of a protein and its various functions in a healthy body—or the malfunction of a diseased one. The hugely complex structure of a protein and the fact that it can change structure in a matter a milli- or even microseconds had made it a very tough problem to crack. But last year, an artificial-intelligence program called AlphaFold, developed by the AI company DeepMind, managed to predict the 3-D structures of almost every known protein—about 200 million in all. This breakthrough is now expected to help revolutionize the healthcare sector across the world.
Generative AI technology that generates deep fakes and writes school essays for lazy students cannot be allowed to act as exemplars of this industry. If the AI industry must take on the current wave of criticism over the impact of AI tools, it should demonstrate use cases that make a real difference to the lives of people at large. It is a good thing going for the AI industry that it has many such case studies. The industry just needs to talk about them a lot more.
Biju Dominic is chief evangelist, Fractal Analytics, and chairman, FinalMile Consulting.
