Viswanathan K's Thoughts

Does the advent of AI mean we need new laws and regulations? When I read two of Rahul Matthan’s posts, I realize they are entirely different questions. The same answer need not apply to both!   Take the laws aspect first . Consider fake news, he says. AI can generate fake content, both audio and video. Does this mean we need new laws? Surprisingly, no. We already have existing laws to cover all this (impersonation of others is a crime; so is harming someone’s reputation). Or consider how AI can be used to trigger communal tensions. Already a crime, covered under the law on using electronic communications to promote disharmony and enmity. Matthan says there is a reason why we can find existing laws for most scenarios: “Pretty much any harm that you think is exacerbated by AI—be it misleading advertising, election interference, forgery or bias—is covered by existing provisions of law. In almost every instance, since these laws have been drafted so broadly as to cover a wide ra...

Some time back, the Titan , a submersible took 5 people to the Titanic wreck 4,000 meters below sea level. As we know, everyone died. Rahul Matthan wrote a piece with an interesting angle to the tragic events.   He writes: “The company steadfastly refused to submit its vessels for classification by third-party agencies like the American Bureau of Shipping.” And passengers had to sign legal undertakings that they understood that the vessel was not certified by any regulatory body; and of course, things could go horribly wrong.   At this point, you’d be shaking your head and muttering about corporate greed and callousness. But wait, there is more. The article is about thinking from first principles. In this case, the first question is – What is the purpose of regulations? Usually, it is to ensure a minimum level of safety and quality.   The company had been claiming that they had far superior and different safety features, but the existing regulatory syste...

Recently there was news that Italy had banned an AI chatbot named Replika. Rahul Matthan’s blog identifies the reason for that – the EU’s GDPR regulation (simplistically put, GDPR is an EU regulation on data privacy). So what are the base principles of GDPR? And what are the problems with those principles when it comes to AI?   Number 1 on that list is “ consent ” – seeking explicit permission before using data. An exception is allowed if said data collection is “necessary” for some “legitimate purpose”. The way AI’s work, they just scour the Net for info and stitch it together in unimaginable ways to derive conclusions from it. By definition, the company that created the AI cannot know to what purpose the AI might put that info to use. GDPR was framed in simpler times when companies could be expected to know what they would do with the data. Not anymore.   Number 2 is that data collection be restricted to what is relevant to the task at hand. And that the data be re...

Recently, the Supreme Court expressed its concern wrt social media and called for guidelines to curb its misuse. While it’s easy to understand the concern, unfortunately there’s probably no solution to the problem. It’s not just about India. Democracies world over make the same noises about social media, but pretty much every time one of them has framed laws on the topic, they’ve solved very little and end up creating a host of new problems. The cure turns out to be the worse than the disease. But why should that be the case? Just because nobody’s framed the right laws doesn’t mean it can’t be done, right? Hmmm… check out Ben Thompson’s podcast on the topic. He starts with the backdrop to social media. Historically, most markets check bad behavior via price. But with the Internet, it’s all free, the tool called “price” became inapplicable as a check. Unlike older mediums, the Internet allows anyone to find like-minded people somewhere or the other. So even if the pe...

For all our lives, there have been rules on how to bring a new medicine to the market: clinical trials, patient consent, double blind trials etc. But all this only started well into the 1900’s, “as drugs became more powerful, so did side effects”, wrote Thomas Hager in The Demon Under the Microscope . And for decades, the way trials were done was, well, horrifying: “(Firms) often went to Africa to do large-scale human tests. In Britain they used soldiers. In the United States, tests were done on prisoners and inmates in mental institutions.” But has the current system of checks and trials gone overboard, wonders Jessica Flannigan in her book, Pharmaceutical Freedom: Why Patients Have a Right to Self-Medicate . In a radio interview with Bob Zadek, she says we should have “freedom over our bodies”, same as the other freedoms we enjoy. People handed the responsibility to the government to prevent a repetition with the “history of drug disasters”. That responsibility became a...

CRISPR, the gene editing tool, allows for the editing of specific genes. The keyword is “specific”. To understand the point, think of breeding racehorses. While one gets to pick the “best” horses to breed, one has no control over which genes the child will inherit. That part’s still a roll of the genetic dice. But with CRISPR, one can target and edit specific genes. The level of control is unprecedented. And scary. Will it lead to the assembly line production of humans “superior” on whatever parameter we choose? Can we anticipate long term consequences of any such changes? Or would it be too late by the time we discover side-effects? Only time will tell how we decide to self-regulate ourselves with such powers. And how successfully those rules are enforced. The one area where it might seem that the benefits would outweigh the risks is when it comes to fixing genetic diseases. But that raises an existential question for pharma and biotech companies, what Goldman Sachs asks i...

When we think of regulations, most of us curse the bureaucrats and politicians responsible for them. Elon Musk, the boss of Space X and Tesla, shares that view. But Musk went a step further: In his biography of Musk , Ashlee Vance quotes Musk’s assessment why regulators rarely change the rules: “If a regulator agrees to change a rule and something bad happens, they can easily lose their career. Whereas if they change a rule and something good happens, they don’t even get a reward. So, it’s very asymmetric… How would any rational person behave in such a scenario?” (As you can see, Musk is hyper-logical. Like Steve Jobs, he demands the impossible from his employees. But Musk, coming from a science background, will also tell his employees that if they can explain to him how the laws of physics don’t allow for what he asks, then he will back down.) But what would happen if we minimized (and in some cases, even eliminated) regulations, outside of areas where people would die e.g. ...

Volkswagen’s “diesel dupe” scandal has been in the news for some time now. 11 million cars worldwide have a “defeat device” that could detect when the car was being tested and then change its performance to improve results. The difference? “The engines emitted nitrogen oxide pollutants up to 40 times above what is allowed in the US.” The CEO has resigned; and the company has set aside $7.3 billion to deal with the problem. Marcelo Rinesi wrote this very interesting article wondering whether the cheating technique used by VW (write intelligent software to behave differently under different conditions) is the ticking time bomb that awaits us in the brave new world of the Internet of Things (In case you are wondering, the Internet of Things refers to the network of objects with sensors and software that can collect and exchange data. So yes, basically that covers almost everything in the future). Rinesi points out the fundamental difference between objects and the new soft...