The topic of generative AI (GenAI) has dominated headlines for much of the last year. A slew of recent breakthroughs has raised concerns among some prominent figures, while the tech industry brims with enthusiasm for what the future might hold.
GenAI also dominated discussion on day one of the 2023 Rakuten Product Conference, which took place on May 31 and June 1, 2023 at Rakuten India’s Crimson House Bengaluru HQ. In addition to a packed audience of Rakuten employees, some 6,000 attendees tuned in from around the world to hear industry leaders share insights on the topic.
Highlights included keynotes from Rakuten speakers such as Group Executive Vice President and CISO Yasufumi Hirai, Chief Data Officer Ting Cai, and Rakuten India CEO Sunil Gopinath. They were also joined by an impressive roster of AI experts from across the industry, including senior director at Adobe Research India Dr. Nanda Kambhatla, Google Cloud India AI ethics expert Kirthi Ganapathy, and many more.
Tech breakthroughs in GenAI
Rakuten’s top AI expert Ting Cai took to the podium to share industry landscapes, opportunities and challenges ahead.
Cai spoke about underlying technology breakthroughs, like evolving from AI developed for specific tasks to today’s general models, the importance of human instruction, and the universal appeal of using natural language to make queries vs. using older-style fixed taxonomies.
“The conversation model used to require understanding specific intent, then it would map it to a task and follow some predefined scripts,” he revealed. “Now the generative AI is a lot more flexible in handling conversations it has never seen before.”
He also shared his views on use cases where generative AI models can make the most impact – specifically, synthesis, creation and automation – and how leaders can make GenAI models accurate, safe and cost-effective.
Dr. Nanda Kambhatla of Adobe Research India highlighted just how transformational GenAI has turned out to be.
“As a machine learning researcher, the whole game has changed,” he told the audience. “It used to be that machine learning meant for every task, for every domain, for every application you want to do, you collect some data, train your model on it, you spend some time finetuning it, and then deploy it for inference.”
Now, massive models trained on vast volumes of data – so-called foundation models – have rendered this process obsolete by becoming flexible, jack-of-all-trades systems that can be adapted for any purpose. “This is becoming the new way of doing AI, and that’s spurring the revolution in generative AI.”
Why the tech industry is so excited
Rakuten India CEO Sunil Gopinath compared the recent breakthroughs in generative AI to the invention of the Gutenberg press in the 15th century and its role in the leadup to the industrial revolution.
“The Gutenberg printing press… made books and knowledge permeable to everybody across the world,” he told the audience. “The generative AI moment reminds me of the same.”
One transformational step in the recent GenAI wave is how major tech players have made their new AI tech accessible to the general public. With tools like ChatGPT, Google Bard and DALL-E, AI is no longer prohibitively expensive for individuals and small businesses to leverage.
“The tech industry stands completely redefined. Generative AI is unlocking value across multiple industries, from material science to creative design, climate change, self-driving vehicles, medical research and education,” Gopinath said. “A commercial ecosystem around ChatGPT and Bard and DALL-E is already developing. We have dozens of startups that already have products in production, whether it’s text, images, video, generating PowerPoints, generating presentations or building spreadsheets.”
“The creativity, the emotion, the empathy, is what human beings are uniquely good at. We hope to leverage AI to augment human creativity and unleash infinite possibilities.”Ting Cai, Chief Data Officer, Rakuten Group
Rakuten also employs AI tools in fintech, digital marketing, customer service, and even medicine to aid research for cancer treatments. Elsewhere in the medical field, Gopinath highlighted a recent AI-enabled breakthrough in the search for superbug-killing antibiotics.
Of course, many more use cases are yet to be fully explored: Cai suggested that GenAI could become a powerful productivity tool as a creative helper in software such as Microsoft 365 or Google Workspace, or as a supercharged learning tool.
But even as AI makes our lives more efficient, Cai argued that humans will still have a role to play: “The creativity, the emotion, the empathy, is what human beings are uniquely good at. We hope to leverage AI to augment human creativity and unleash infinite possibilities.”
New challenges in ethics
Even with its many remarkable applications, Gopinath acknowledged that some experts are rightfully concerned by the rapid evolution of GenAI. “[Yuval Noah] Harari, one of the most celebrated historians in the world today, has said that AI has hacked the operating system of human civilization.”
Other prominent voices have raised questions surrounding privacy, data bias, and safety, as well as around how GenAI might effect the creative industries. Governments, meanwhile, are watchful about the potential of deepfakes to misinform people.
“How do we come together as an industry and address these issues? Who owns content? How do you solve for bias, particularly with enormous amounts of training data, on a very narrow set of humans or use cases? How do we solve for safety for once AI gets its own mind? Who designs the rules of how AI is controlled? These are the fundamental challenges we as an industry need to come together and address today.”
Ethics expert Kirthi Ganapathy of Google Cloud India stressed the importance of tackling ethical issues before they become problematic.
“Ethics can no longer be in retrospect of action,” she told the audience. “It has to be preemptive. This is where a human in the loop becomes essential. In order to have rightful control of what kind of responses are coming out before things get out of control. At the end of the day, ethics is a human function.”
The impact of AI on some human jobs – another parallel to the industrial revolution – is another challenge that must be addressed with empathy. Gopinath referenced a report from Goldman Sachs that suggested some 300 million jobs might be affected by generative AI in the coming five years.
Gopinath told the crowd that as a tech company, Rakuten must take a leadership role in this emerging field: “AI for us is not an outcome. It’s not optional for our scale. AI is our bread and butter. It’s our culture,” he said. “But it has to be tempered with our responsibility to our society. How do we do it in an empathetic way?”
Share the knowledge, and stay human
Rakuten Group Executive Vice President and CISO Yasufumi Hirai offered valuable perspective on the everchanging nature of technology itself.
“We are shifting our world from an industrial society to a knowledge society. The industrial society was built on the rule of distribution – distribution of natural resources, distribution of labor, distribution of wages,” he explained. “But the knowledge society is built on the sharing concept. One hammer cannot be shared with two persons, but knowledge can be shared immediately by millions of people. That’s the society we are living in right now.”
Generative AI, he argued, will play a significant role in taking our knowledge society to the next level – but only if we can remember who the technology is supposed to help.
“I am concerned that many tech engineers start their sentences with technology as the subject. But the subject of any sentence should be people; society; the world; organizations; family,” he stressed. “Technology can be the enabler to realize something amazing.”