- Is AI about to take your job? It depends on the outcome of a heated battle in Silicon Valley.
- Goldman Sachs estimates 300 million jobs could be affected by AI.
Is AI about to obliterate all our jobs? Depends on who you ask.
A new report from Goldman Sachs reckons generative AI models like OpenAI's GPT-4 could impact 300 million full-time jobs globally. That means around 25% of the current workforce could be substituted by AI, while two-thirds could see some degree of automation to their jobs.
But whether those workers are displaced fully or find new roles seems increasingly to depend on who turns out on top in a raging AI war taking shape in Silicon Valley — notably between OpenAI and Google.
Sam Altman, CEO of the ChatGPT-maker OpenAI, has admitted being scared of his own creation, warning that it could "eliminate a lot of current jobs" (though he optimistically thinks new ones will replace them.) Satya Nadella, Microsoft CEO and Altman's newfound partner, reckons AI will make existing workers more productive.
Google's Sundar Pichai has gone as far as calling AI "more profound than fire or electricity" — a technology that needs to be regulated.
These differing views are not mutually exclusive, but demonstrate the comparative lack of guardrails around AI right now.
Earnings calls are scattered with references to AI. Investment in the tech is skyrocketing as CEOs seek to reap the advantages on offer — possibly at their human workers' expense.
The technology has evolved so quickly that regulation hasn't kept up. As it stands, humanity's future employability is dependent on whichever AI company wins out behaving responsibly. And even their own employees, investors, and peers aren't convinced.
OpenAI is effectively dictating the speed of AI rollout
It's pretty clear that OpenAI is in the lead — and it's less fettered by ideas of safety than its competitors.
When the firm released ChatGPT in November, it almost immediately disrupted education and hiring, as students and jobseekers used to tool to write essays and applications. The next iteration, GPT-4, is even more advanced and makes a reasonably competent programmer, philosopher, lyricist, and therapist.
Google has recognized its need to keep up and released its own AI chatbot Bard to the public last week. But the technology isn't as obviously powerful as GPT-4, and has left early users underwhelmed. The company has also been publicly much more cautious about launching AI tools in a "responsible way," in part because it has more to lose reputationally if its releases go wrong.
OpenAI's apparent willingness to roll out advanced AI models as soon as they're ready points more to a future where the tech replaces vast swathes of the workforce. Coinciding with all of this is the biggest jobs contraction in the tech industry's history, as leaders look for cost savings and productivity gains.
AI will sure look like a cost saving
Goldman's analysts wrote in their jobs note that "the boost to labor productivity growth could be much smaller or larger depending on the difficulty level of tasks AI will be able to perform and how many jobs are ultimately automated." The smarter the AI, the quicker jobs will be replaced.
They estimate productivity growth of around 1.5% in the US labor market over the next 10 years, assuming widespread AI adoption. In one modeling example, they estimated that on a difficulty scale of 1-6, where 6 is the highest difficulty, an AI model capable of performing a level 6 task, such as analyzing "the cost of medical care services for all US hospitals", would help boost productivity by 2.9% per year.
Given the evolution of large language models in recent months, it seems likely that AI is getting closer and closer to accomplishing difficult tasks alone.
While optimism is high and oversight is minimal, OpenAI has every incentive to push ahead with monetization while it has mindshare and (publicly at least), the most impressive tech. To that end, the firm has already rolled out a subscription version of ChatGPT. By the time regulators step in, it will be difficult to undo enterprise AI adoption.
History dictates that innovation with the power to disrupt and eliminate jobs typically paves the way for new fields, in time. A 2020 report from the World Economic Forum predicted AI will displace 85 million jobs by 2025, and create 97 million new ones.
What seems different, this time, is the step-change in the power of the technology at play. Generative AI is proving itself to be anything but a prosaic tool that makes life a little bit easier for users. It is showing the world it can rival humans in all sorts of fields. And until lawmakers step in, humans may be at risk.