GPT-4o, the company’s newest multimodal model, can input and output a combination of text, audio, and images. The technology represents a major advancement of the artificial intelligence of the recent past.
OpenAI announced the model in a series of demo videos on Monday, showcasing the technology’s improved vision and voice abilities. The videos elicited both wonder and mockery, with people quickly making comparisons to the 2013 sci-fi movie “Her” and Elon Musk saying the reveal made him “cringe.”
While it’s too early to predict how exactly the model will disrupt the workforce, GPT-4o and other multimodal models will inevitably change the way we work, said Maribel Lopez, an AI analyst and founder of research and strategy consulting firm Lopez Research.
“The concept of multimodal models will impact a lot of different industries because it handles text, video, and audio,” Lopez told Business Insider.
But not all of those impacts will necessarily be negative, Lopez said. For example, electricians, plumbers, and other tactile workers could use multimodal AI to make their jobs easier, she said.
“For workers who fix specialized equipment, AI might be very helpful in troubleshooting or fixing problems,” Lopez said. “But it won’t replace them because they have to be there to do it.”
While some companies are working on AI robots that can do physical work, those models are typically better suited for repeated menial tasks like welding bolts than complex blue-collar labor.
However, other industries could have more of a challenge adapting to the implementation of multimodal AI in the workplace.
“AI will impact any job that has data,” Lopez said, pointing to industries like supply chain and finance.
The general consensus is that anywhere from 20% to 30% of tasks completed by “computer workers” will eventually be offset by AI, Lopez said. But that doesn’t mean computer workers will find themselves unemployed.
Using paralegals as an example, Lopez said their jobs could shift from tracking down documents and writing up summaries — two tasks that can take a person hours but which AI can complete in minutes — to tasks yet unknown.
“The AI challenge is that it forces all of us to enhance our skillsets,” Lopez said. “It will be a change for all of us.”
And it appears that OpenAI doesn’t want GPT-4o to feel like doomsday for computer workers, either.
The company included a demo video showing the technology acting as a personal assistant, suggesting changes to code in real time, and offering a one-sentence summary of text.
