To ensure their safety, open-weight models are being released further and further behind the current state of the art. What does this mean for Indian AI developers who have traditionally preferred open-weight models for the flexibility they offer?
A consensus seems to be emerging around the world that what we need to do to protect people is label AI-generated content so that they are not misled into thinking it is real. The trouble is that watermarks are easily circumvented. Would we not be far better off just calling out what little content is real?
There is no doubt that AI is a technology like no other and that, as a result, the disruption it is likely to cause will be unprecedented. But despite the existential fear that this has brought about, it is unlikely that it will render us irrelevant. Not so long as we continue to constantly question.
_If India wants to be a significant player in the AI space it will need to take a concerted effort to actively encourage investments into the space. The good news is that we have done this before. All we need to do is adapt our IT playbook to the AI age.
The report of the RBI Committee on the Framework for the Responsible and Ethical Enablement of AI recommends a novel approach to AI regulation - one that encourages innovation and yet mitigates risk. It offers a blueprint for AI regulation in India.
Even though it is framed as an author’s right, copyright law was originally designed to protect publishers. This is why, to this day, authors rarely make enough from their books to earn a living. What authors want, most of all, is to be read. If AI does that, shouldn’t we rethink copyright?
Laws have always evolved in response to technological change. We’ve had a long tradition of this sort of change ever since the Industrial Revolution. We are standing on the threshold of a similar technological shift, and our legal system is likely to change yet again.
It might appear as if AI is ingesting copyrigjht material - text, images, videos, etc. - and regurgitating it out in response to prompts. But if you peel back the layers, the underlying process looks a lot more like learning than copying.
To withstand scrutiny, a forgery must extend beyond the artefact in question to the entire back-story within which it resides. While it used to be that these narratives had to be concocted painstakingly, one instance at a time, modern technology makes it possible to carry out on an industrial scale.
We have been trained to think that all bias is bad. And that we should do all we can to rid ourselves of it to avoid being unfair. What we often fail to realise is that bias is a shortcut that allows us to cut through the noise and zero in on the data we need to process.
