Recently, the Wall Street Journal (WSJ) expressed a skeptical view on the current state of AI technology, claiming that “the AI revolution is already losing steam.” This has left many people curious, so today, let’s delve deeper into this issue.
AI Innovation Slowdown: Is It True?
First, the WSJ claims that the pace of AI improvement is slowing down. Large language models (LLMs) like OpenAI’s ChatGPT and Google’s Gemini have improved their functionalities by learning vast amounts of text. However, it is now said that there is a lack of additional data for further learning. AI researchers are trying to resolve this by using ‘synthetic data’ generated by other AIs, but the effects have been minimal.
Cognitive scientist Gary Marcus stated, “AI like ChatGPT improved rapidly initially, but recently, we have only seen incremental increases.” His diagnosis is that the core functionalities of AI have either stabilized or at least the pace of improvement has slowed down.
Investment Profitability of AI: An Uncertain Future
Investment profitability in AI is also a significant issue. Anshu Sharma, CEO of AI startup Skyflow, said, “The future of AI startups is bleak.” For AI startups like OpenAI and Anthropic to succeed, they need to invest massive amounts of capital over a long period, which is realistically challenging.
In fact, the CEO of Stability AI, the creator of the popular image generation AI tool ‘Stable Diffusion’, suddenly left the company. Many other AI startups are also in discussions for acquisition.
High Operating Costs of AI
Operating costs of AI are also problematic. AI needs to think anew each time a user asks a question, and the resources required to generate answers are much greater than traditional searches. This could potentially erode the margins of companies using AI technology.
For example, Silicon Valley venture capital firm Sequoia calculated that the AI industry spent $50 billion buying chips from Nvidia to train AI, but only made $3 billion in revenue. This showcases how burdensome AI operating costs can be.
Limited Increase in AI Usage
According to a survey by Microsoft and LinkedIn, 3 out of 4 office workers currently use AI at work. However, this is still far from the numbers AI companies need. Peter Cappelli, a professor at the University of Pennsylvania’s Wharton School, said, “AI has not been as helpful in improving productivity as it was advertised.”
Conclusion
Is AI really our future? The WSJ did not completely deny the potential of AI, stating, “We cannot say that AI will not change all kinds of jobs and industries in the long term.” However, the current evidence suggests that the belief in AI’s rapid development and adoption is questionable.
We need to adjust our excessive expectations of AI and reassess its role from a realistic perspective. It is crucial to move away from the illusion that AI will solve all problems and find ways to use AI correctly.
Source: WSJ, “The AI Revolution Is Already Losing Steam”