Artificial Intelligence at Your Command

RIGHT UNTIL THE end of last year, I thought Artificial Intelligence (AI) was more fad than promise, my view being that it was programming by another name, just more computing power backed by more data and networks. But that wasn’t anything novel. What a mainframe did in the 1980s, the mobile phone can do now. The debate goes on endlessly, ranging from whether AI even exists to whether human intelligence exists because, after all, the brain too is a computation machine that uses a different mechanism. Even so, an intellectual argument changes somewhat when you see what exactly AI, even if just a label, can do. I found this out when a cousin of mine dropped in at the beginning of December. He had studied animation in the US and was working on special effects in a film production company in Mumbai. He casually mentioned being afraid of not having a job in the future because of what AI was doing to art. I asked him how it works and he pulled out his iPad and opened a program called Midjourney and asked me for a description. I made it as wild or impossible as I could imagine. I don’t remember the exact line that I came up with, but in a matter of seconds an image of that line had appeared in front of me. It looked like something years of art school would be needed to produce and, even though I couldn’t draw a straight line, the artwork was entirely mine. Later, I got on to another AI art system called DALL.E, which is the most popular one around, and I asked it to come up with the image of a man standing on the edge of the solar system looking into a black hole in the style of the artist SH Raza. Within seconds, there was a silhouette looking down a hole of concentric lines in what typifies Raza. I asked it to do MF Husain’s self-portrait in the style of Vincent Van Gogh and it came out with a decent impression. My cousin was right. It had all the potential to put him out of a job because there would be no skill needed anymore to be an artist; just imagination was enough.

I still don’t think AI is an extraordinary phenomenon jumping out of nowhere to change humanity. Technology moves incrementally but what was certain was that a tipping point had arrived. Or, at least we can now see what the tipping point is going to be. If 2022 was when we got that notice, 2023 will be when the usage goes up and not just for those who go searching for it. That is because most of what we see now are by entities that are not part of the big technology companies. ChatGPT was the other major AI system that made waves last year. It could answer any question you threw at it in the way a human being would. Both ChatGPT and DALL.E are products of OpenAI, a quasi-commercial entity whose main motive is not profit but steering AI into the future. Making their systems available to the public is pressure on the Big Tech companies, which have also been working on their products, to do likewise. ChatGPT, for instance, is now touted as a threat to Google Search and that is a forced incentive for Google.

In December, Google had a meeting of its employees where they asked the management about plans to counter ChatGPT and whether they had missed the bus given that they have had a similar product called Lamda in the works. Reporting on it, CNBC wrote: “Alphabet CEO Sundar Pichai and Jeff Dean, the long-time head of Google’s AI division, responded to the question by saying that the company has similar capabilities but that the cost if something goes wrong would be greater because people have to trust the answers they get from Google. Billions of people across the globe use Google’s search engine, while ChatGPT just crossed 1 million users in early December.” But Pichai also “said at the meeting that the company has “a lot” planned in the space for 2023, and that “this is an area where we need to be bold and responsible so we have to balance that.”…Pichai said that 2023 will mark a “point of inflection” for the way AI is used for conversations and in search. “We can dramatically evolve as well as ship new stuff,” he said.”

ChatGPT and DALL.E come under the umbrella of generative AI—you ask it to create something based on what you want, the former does words and the latter images. Stanford University’s Institute for Human-Centered AI asked its faculty to make their predictions on what AI held for 2023 and one of them, Percy Liang, said that this would be the year of generative video. He was quoted on their website: “We may be getting to a point next year where we won’t be able to distinguish whether a human or computer generated a video. Up to today, if you watch a video, you expect it to be real, but we’re seeing that hard line start to evaporate.” It doesn’t need much to imagine its impact on social media. Creators on platforms like TikTok and Instagram’s Reels would flock to it. Just as DALL.E makes anyone an artist, generative videos will make everyone a filmmaker at some point in the future. One of the big players in this area is Facebook. They already announced a product called ‘Make A Movie’ last September and while still bare bones and without access to the general public, they too will need to enter the market as smaller startups get on the generative AI field. The quality of the video generated will be clumsy in the beginning but the future is limitless. Facebook’s AI program, for instance, is designed to be part of their ambitious Metaverse, or virtual reality universe, that Mark Zuckerberg is envisioning.

IN THE STANFORD article, another faculty member made a prediction for the year that wasn’t rosy. Russ Altman, an associate director of their Institute for Human-Centered AI, felt that too much would enter too soon and it would inevitably mean abuse. The adjective he used for the imminent rollout of the technology was “shocking” because it was not mature or regulated enough. The article said: “He points to the recent proposal in San Francisco to allow police to deploy potentially lethal remote-controlled robots or the potential misuse of tools that can generate human-like text from a short prompt—think how many smart fifth-graders could skip an essay assignment by asking an agent for help, he says. For 2023, “I expect a hit parade of AI that’s not ready for prime time but coming out because it’ll be driven by over-zealous industry,” Altman says.”

Writing in Forbes, venture capitalist Rob Toews, who invests in AI, argued that conversational search would be one of the online gamechangers in 2023. Instead of links in response to a query, the information would become more palatable as part of a conversation with an AI tool. At present it is not accurate, the reason that Pichai cited for not making haste but Toews still felt that search would change in the biggest way since Google first revolutionised it. He had altogether 10 AI predictions for 2023 that included humanoid robot launches, driverless cars being normalised, and an uptick in the research usage of Alphafold which can predict 3D models of protein from amino acid chains that has enormous potential to develop better medicines.

I asked ChatGPT to write a poem on “AI in 2023” in the style of the great Welsh poet Dylan Thomas and it came up with: “In 2023, AI did reign supreme/ It seemed to know just everything/It could solve problems with such ease/It seemed to have an endless spring.” Just about everything in it was wrong. It read like a schoolboy’s attempt at rhyming and Thomas would be mortified at his name being associated with something that has no connection at all to how he wrote. The ideas the verse speaks of like AI knowing “just everything” or “reigning supreme” was self-evidently flawed. And when I asked it to write the same poem in the style of TS Eliot, it wrote in the same rhyming vein as earlier: “In 2023, the machines did rule/Their algorithms cold and hard/They seemed to know just what to do/Their logic impossible to bard.”

What this showcase of its creative prowess tells us is that it presently doesn’t really care what the answer is so long as it gives an answer. The year will see many beginnings in AI but, if a poem written by it is any indication, it will still be a work in long progress.

About The Author

Madhavankutty Pillai

•