Tech Tonic | The legacy of Mira Murati, and more chapters remain to be written

Tech CEOs often get the credit, but it is the groundwork led by CTOs such as Mira Murati, which defines a company’s fortunes and indeed AI as we know it today.

The exodus at OpenAI is going from bad to worse. Or so it may seem to us, observing from the outside. It is believed that much of the team that worked on safety matters related to artificial general intelligence (AGI) left a few weeks ago. Now it is the turn of three leaders. Chief research officer Bob McGrew and vice president of post-training Barret Zoph, alongside perhaps the most popular OpenAI exec after Sam Altman, the company’s now former chief technology officer, Mira Murati. Concerning, to say the least.

Mira Murati

It is Mira Murati’s legacy, which in my opinion still has many a chapter yet to be written, that we must talk about. The artificial intelligence (AI) landscape as we know it, is mostly due to her doing all that she did in the six years or so at OpenAI. There is no need to beat around the bush—Murati is responsible for the launch of ChatGPT, the text-to-image generator DALL-E and has no doubt played a key role in piecing together the foundations of the upcoming AI video generator, Sora. Days earlier, OpenAI announced the Advanced Voice Mode for ChatGPT Plus subscribers, for what they claim are natural, real-time conversations.

The AI landscape that you are so used to now, and all those hours spent chatting with ChatGPT, you can thank Murati for her role in making those a reality. And it is not just about how we interface with AI. These are the very products that have slingshot OpenAI to the very front of the AI race, leaving tech giants including Google, Meta and Microsoft (their investment in OpenAI is intriguing) scampering to keep pace. OpenAI no longer has the sort of lead it probably had a year or so ago, but its tech continues to leave little to chance.

The last 12 months at OpenAI have been tumultuous, but that didn’t stop significant updates for the GPT and DALL-E models (including the flagship GPT-4o) through this time. So much so, Murati teased that one of the upcoming GPT models (perhaps early 2025) will have “PhD level intelligence”. Even if not, that’s a statement of intent. One of confidence. That could very well be the reason Apple finally decided to have GPT as the underlier for a smarter Siri that rolls out in the next few months as part of the phased Apple Intelligence rollout on iPhone, iPad and Mac.

Murati’s never been far from a controversy or two. She didn’t have a clear answer as to where OpenAI’s video generation model Sora’s training data comes from. Perhaps a refined art of the poker face may have helped at that point but it is clearly a skill she doesn’t have (among the many she has).

I had written, after an inevitable social media firestorm when she predicted some creative jobs may be replaced by AI in the coming years, that she’s said what she’s said, based on a broader view of how AI is being built in research labs. A privilege most of us don’t have, and certainly not the keyboard warriors on X and wherever else social media boffins tend to be found. Before OpenAI, she worked at Tesla, and that timing coincided with the initial versions of the Autopilot AI driver assistance software.

And then at Leap Motion, a company that wished to replace the keyboard with augmented reality. An idea, much before its time. I remember using a Leap Motion integration in one of HP’s laptops (that must have been early 2014, if memory serves me well).

Murati has often said that the public, that is the users, should play a role in defining the path of AI. OpenAI’s products over the years have tried that method to tangible success. Google’s Bard (and subsequently being renamed to Gemini) and X’s Grok have tried learning in the real world and stumbled. Murati’s vision for building AGI with the public as part of that conversation, as opposed to building this in the confines of labs, now stares at an uncertain future. We don’t know OpenAI’s next stance, for now.

What happens to Sora’s development? Will Murati’s stamp be visible on that generation tool when it finally rolls out to users? Or will Adobe’s own Firefly-based model, which they talked about a few weeks ago, pip it to the finish line?

We don’t, at least not at the time of writing this, know Murati’s next destination. Smart money would be on her ending up at Apple (there may be a contractual clause or two, based on time), perhaps even Google. That is, unless she has her heart set on an AI start-up—OpenAI’s Ilya Sutskever has walked that path already.

As Murati steps away from OpenAI and everything she helped build in the past few years, her legacy is one of being the leader who got us to where we are, with our interactions with AI. She often gets little credit for it (CEOs do, and that’s also got more to do with airtime), but it’s the tech she’s helped build, that has had an undeniable impact. I can say with a degree of certainty, she’s not done just yet.

Vishal Mathur is the technology editor for the Hindustan Times. Tech Tonic is a weekly column that looks at the impact of personal technology on the way we live, and vice-versa. The views expressed are personal.