Open AI Innovation

All innovations build on those that came before. It is important to strike a balance between leveraging intellectual property protection for commercial benefit and stifling innovation in the process. And this is particularly relevant in the context of AI.

This is a link-enhanced version of an article that first appeared in the Mint. You can read the original here.

Towards the end of 1901, Guglielmo Marconi set out to conduct an ambitious experiment. At the time, it was believed that since electromagnetic waves (like light) travelled in straight lines, radio waves could not be used to transmit messages across large distances—much like how even the most powerful telescope could not catch visual signals from places located beyond the earth’s horizon.

Marconi set out to prove them wrong. He installed a massive (70-metre high) cone antenna on the Poldhu Cove in Cornwall, UK, from which he planned to transmit a message in morse code to receiving equipment he had set up at Signal Hill in Newfoundland on the other side of the Atlantic Ocean. At 12:30pm on 12 December 1901, Marconi heard three faint but distinct clicks over his headphones at the receiving station at Signal Hill (corresponding to the letter ‘S’ in morse code). With that he proved that it was, in fact, possible to communicate wirelessly across the Atlantic Ocean, and in doing so, launched the modern telecommunication revolution.

Even though he is widely credited as the inventor of radio communication, Marconi’s Grand Transatlantic experiment would not have been possible without the contributions of Jagdish Chandra Bose, one of India’s finest scientists. In 1899, Bose invented the ‘iron-mercury-iron coherer,’ the first device that was capable of continuously detecting radio frequency signals. While coherer technology had been around for a while, Bose’s coherer was the first that was self-restoring, allowing for the continuous detection of radio signals without the need to manually reset the device after each signal is detected. This continuous signal processing capability remains, to this day, a critical feature of modern mobile communication devices.

Bose refused to patent his invention, firm in his belief that scientific knowledge must be freely exchanged. It was as a direct result of this philanthropic approach that Marconi was able to incorporate that coherer design into his Signal Hill receiving station. The rest, as they say, is history.

Marconi was not, however, of a similar mind. Not only did he refuse to acknowledge Bose’s contribution, he intentionally obfuscated the details of how he came upon his coherer technology. From the moment it was clear that his experiment was a success, he understood the commercial significance of the radio communication technology he had invented. From that moment on, he not only took great pains to secure all intellectual property rights over his invention, he aggressively enforced them in various ways.

He refused to sell his equipment to customers, insisting that they lease it from him instead. What’s more, even after leasing equipment from him, customers could only use operators under Marconi’s employment to operate the equipment. He also insisted that transatlantic vessels on which his radio equipment was installed could only communicate with other similarly fitted vessels, further cementing his monopoly by refusing to allow any other competing technology to inter-operate with his.

Given how focused he was on short-term commercial gains, the fact that these measures would hinder broader scientific research in radio technology did not seem to matter to him. As a result of this, however, innovation in radio technology stalled, remaining restricted to spark-gap technology until about 1912, even though continuous wave technologies offered lower interference, more efficient use of radio spectrum and the ability to carry audio information, enabling voice and music broadcasting.

New inventions always build on the innovations that came before them. Continued scientific progress thrives in an environment of open scientific collaboration.

While intellectual property laws offer inventors some sort of temporary legal monopoly over their ideas for long enough to recoup their investments in research, over the long-term scientific progress will only take place if we can all collaborate in a scientifically permissive environment. Unless we enable this sort of an approach, it will be impossible for us to conduct inter-disciplinary research where experts from different fields come together in order to tackle larger and more complex problems.

The discussion around open innovation has, in recent times, resurfaced in the context of artificial intelligence (AI). Most of the large language models available for use today are proprietary. Information about these models and their training are treated as a closely guarded secret by the companies that have created them. Leading AI companies and investors defend this approach, claiming that AI is a dangerous technology that needs to be controlled, given the harm it can cause.

While it is possible that some of these fears may be true, I cannot help but think that it is a bit self-serving. As if, having secured a first-mover advantage, these companies are looking to pull the ladder up behind them by insinuating that they and only they have the technical knowledge required to keep the world safe from the harms that AI can cause.

Be that as it may, history has shown that no good comes of limiting access to new technologies and inventions. In fact, to the contrary, every time we have opened up new and seemingly dangerous technologies by letting their innards into the public domain, the more we’ve benefited from the many ways in which they are used.