Imagine having a smart AI assistant that can handle your daily tasks without you having to lift a finger—scheduling meetings, sending emails, or even online shopping. That's the idea behind OpenAI's latest innovation, Operator. Instead of just providing text-based answers, the AI model, scheduled for release in January 2025, can directly control computers and handle complex tasks, including booking tours, browsing the web and writing code. The user marks a major step in “agent AI,” autonomous systems that can take action and complete tasks without human supervision. OpenAI CEO Sam Altman hailed agent AI as the “next big breakthrough” of the time recent Reddit AMA.
“AI-powered AI systems can store the user's context and memory, making communication natural and continuous without the need to retrieve information,” Yang Yuanqing, CEO of Lenovo Group, told the Observer last month during an interview at the Lenovo Tech World event in Seattle. . “These autonomous systems create a unique intelligence base within consumer devices that can adapt to individual needs and improve productivity without requiring any operating technology to operate.”
Lenovo is one of the many tech companies joining the race. The PC maker recently announced “AI Now”, an AI agent system built into Meta's Llama 3.1 that promises to make PCs smarter by turning them into personal digital assistants. AI can now automatically edit documents, schedule meetings and adjust device settings with voice and touch commands—all without requiring an Internet connection.
“The ability to store the context of agent AI systems allows us to deliver a richer user experience,” Yang said. “Our AI model learns from individual behavior and choices while securely storing this information on the device to ensure privacy and compliance.”
OpenAI's direct competitors, including Anthropic, Microsoft, Google (GOOGL) and Salesforce, are also racing to develop next-generation AI agents.
Microsoft recently introduced its autonomous AI agents to the enterprise, integrated with its Dynamics 365 platform, aimed at automating tasks across the sales, finance and supply chains. Salesforce launched Agentforce, which enables businesses to create customized AI agents for marketing, customer service and IT. Anthropic, on the other hand, introduced Claude 3.5 Sonnet, an AI agent that can control computers and perform advanced coding and computing tasks.
“The rise of agent AI in fields like IT is pushing organizations to adapt,” Shruti Dhumak, senior customer engineer at Google, told the Observer. “Soon we will see a shift from people mainly doing tasks to planning them, working with AI agents to tackle complex objectives. Companies will need to invest heavily in AI talent and perhaps rethink their team structures, with new roles focused on human-AI interactions emerging. “
Can “agent AI” take away human jobs?
With great power comes great disruption. AI agents may take on roles once reserved for humans. “It's going to have an impact initially on the job market, where low-skill, repetitive tasks such as the completion of customer service,” Nish Krishna, founder of AI development platform Fractionalize, told the Observer. “Even roles that require intermediate skills, such as planning and other health care positions will feel the effects.”
Krishna warned that the short-term effects could be severe, especially for older workers, some of whom may find it difficult to return to work. “The workplace could be impacted as companies spend more time in the coming years rethinking the responsibilities that AI agents use,” he said.
Some experts say it's too early to worry. “The idea of fully autonomous AI, as promoted by leaders like Sam Altman, ignores the vast amount of human resources needed to implement these systems,” Annie Brown, an AI researcher at UC San Diego, told the Observer.
Peter Wang, chief AI innovation officer and co-founder of Anaconda, the world's largest data science tools provider, believes that, while the proliferation of AI agents may make some jobs obsolete, it will also create new opportunities. “The dynamic nature of the workplace will move from a person-to-person interaction to a person-to-agent relationship,” he told the Observer.
To stay competitive, workers must learn to use AI tools in specific jobs and industries, and businesses must invest in continuing education to help workers adapt, Wang said. “This investment can include continuing education programs that combine technical skills, such as data literacy and API integration, with creativity and problem solving.”
