'We Think It's Overhyped': AI Is in For a Humble Reality Check in 2024, Analysts Say Analyst firm CCS Insight predicts the generative artificial intelligence (AI) space is poised for a "cold shower" in 2024.
By Madeline Garfinkle •
Key Takeaways
- Developing and maintaining generative AI is becoming expensive for organizations, leading to concerns about the financial sustainability of AI projects.
- The growing computing power required for AI comes with significant environmental implications.
Opinions expressed by Entrepreneur contributors are their own.
Despite artificial intelligence's sweeping presence and mainstream implementation, this year's hot tech buzzword could be in for quite the humbling by next year.
According to a recent prediction by analyst firm CCS Insight, the generative artificial intelligence (AI) space is poised to get a "cold shower in 2024" as escalating operational costs and increasing demands for regulatory oversight "replaces the hype," CNBC reported.
"We are big advocates for AI, we think that it's going to have a huge impact on the economy, we think it's going to have big impacts on society at large, we think it's great for productivity," Ben Wood, chief analyst at CCS Insight, told the outlet. "But the hype around generative AI in 2023 has just been so immense, that we think it's overhyped, and there's lots of obstacles that need to get through to bring it to market."
The cost of deploying and maintaining generative AI is "immense," and it will "become too expensive" for many organizations and developers, Wood added. AI models like ChatGPT or Google's Bard rely heavily on high-powered chips like Nvidia's graphics processing units (GPUs), which come with a price tag of roughly $10,000 each — and that doesn't even account for the cost of maintaining the program.
Running ChatGPT, for example, costs OpenAI about four cents per user query, according to estimates by Bernstein analyst Stacy Rasgon, per CNBC, and if queries were to reach a tenth of the scale of Google search, it would translate to an initial investment of approximately $48.1 billion in GPUs, along with an annual expense of $16 billion for chips to maintain its operation.
As a result, companies like Amazon, Google, Meta, and OpenAI are developing their own AI-specific chips to support their AI workloads and minimize the cost.
Furthermore, the computing power necessary for generative AI comes with environmental caveats. A recent peer-reviewed analysis conducted by Alex de Vries, a data scientist and Ph.D. student at Vrije Universiteit Amsterdam, presents early estimates on the tech's environmental impact, projecting that by 2027, AI servers may use between 85 to 134 terawatt hours (Twh) annually.
The hype around generative AI in 2023 has just been so immense, that we think it's overhyped, and there's lots of obstacles that need to get through to bring it to market.
The energy usage is equivalent to the annual electricity consumption of countries like Argentina, the Netherlands, or Sweden, and roughly 0.5 percent of the world's current electricity use, the New York Times reported.
The environmental impact of AI's electricity demand depends on the energy source of data centers, whether they rely on fossil fuels or renewable resources. In 2022, data centers supporting computers accounted for about 1 to 1.3 percent of global electricity usage, excluding cryptocurrency mining, which added another 0.4 percent, according to sustainability organization, IEA.
"Maybe we need to ideally slow a bit down to start applying solutions that we have," Roberto Verdecchia, an assistant professor in the University of Florence's Software Technologies Lab, told the Times. "Let's not make a new model to improve only its accuracy and speed. But also, let's take a big breath and look at how much we are burning in terms of environmental resources."
Along with other factors contributing to AI's hype decline, CCS Insight predicts that AI-based identity fraud will increase in 2024 as individuals use the tools to impersonate others through deepfake technology or voice synthesis.
