Ed Zitron, host of the Better Offline podcast and writer of the insightful newsletter "Where's Your Ed At?", dives into the competitive AI landscape with Brooke Gladstone. They discuss how China's DeepSeek R model has emerged as a formidable player, produced at a fraction of the cost and time of U.S. counterparts. Zitron critiques OpenAI's financial struggles despite high revenues and highlights the industry's obsession with trends like the metaverse, questioning the sustainability of current tech models and the risks posed by monopolistic practices.
DeepSeek's cost-effective approach to AI development challenges the prevailing notion that high investment and resources are necessary for success.
The scrutiny surrounding generative AI's practical reliability raises concerns about the long-term viability of existing models and technology in the industry.
Deep dives
Disruption in AI Industry
The introduction of DeepSeek, a Chinese AI model, has significantly disrupted the AI industry, competing with established giants like OpenAI and Google DeepMind while being much more cost-effective. DeepSeek managed to develop models comparable to GPT-40 for a fraction of the cost, demonstrating that expensive resources like high-performance GPUs are not the only path to success in AI development. This advancement has caused stock prices of major tech companies to plummet almost instantly, with NVIDIA reportedly losing billions in market value in one day. The open-source nature of DeepSeek's technology means that anyone can adapt and build upon their models, further accelerating competition and innovation in the sector.
Revelations About AI Development
The AI industry has long been perceived as requiring massive investments in resources, as exemplified by the enormous capital behind companies like OpenAI and Anthropic. However, DeepSeek’s approach has revealed that efficient modeling and training can be achieved at a fraction of the cost, using only a small fraction of the GPUs previously thought necessary. This has led to the realization that the benefits of entrenched companies may have been based more on narrative than on actual technological necessity. By publishing their methods and research, DeepSeek has opened the door for others to innovate without the burden of monumental costs.
Challenges of Generative AI
Despite the initial hype surrounding generative AI, its practical applications have come under scrutiny, particularly when it comes to reliability and effectiveness in real-world applications. Critics argue that large language models, including those developed by OpenAI, often fail to deliver accurate and coherent results, especially in demanding scenarios. This inconsistency raises questions about the fundamental capabilities of generative AI and the effectiveness of the solutions proposed by big tech companies. As these models are built on probabilistic patterns, they inherently lack the ability to understand the context or semantics of their output, perpetuating issues such as hallucinations—instances where the AI presents false information as true.
Implications for the Future of AI
The emergence of DeepSeek signals a turning point in the AI landscape, highlighting the potential for a significant downturn in the tech industry if the anticipated growth does not materialize. There is a fear that the current AI bubble is primed to burst, with major companies facing pressure as they struggle to demonstrate sustainable profitability amid rising competition. Many experts believe that without genuine innovation or clear utility, the long-term viability of the existing models is in jeopardy. This scenario may lead to mass layoffs in the tech sector similar to those seen during past economic downturns, alongside a reevaluation of how AI companies are valued in the marketplace.
When OpenAI launched ChatGPT in November 2022, all the big tech firms were clamoring to make their own versions of the “intelligent” chatbot. Billions of dollars have been thrown into the technology – training the models, creating more advanced computer chips, building data centers. But last week, a Chinese artificial intelligence company called DeepSeek released a generative AI model that is not only competitive with the latest version of OpenAI’s model, but it was done cheaper, in less time, and with less advanced hardware. For this midweek podcast extra, host Brooke Gladstone sat down with Ed Zitron, host of the Better Offline podcast and writer of the newsletter “Where’s your Ed at,” to talk about how this new Chinese AI model threatens to burst the American tech AI bubble.
On the Media is supported by listeners like you. Support OTM by donating today (https://pledge.wnyc.org/support/otm). Follow our show on Instagram, Twitter and Facebook @onthemedia, and share your thoughts with us by emailing onthemedia@wnyc.org.
Get the Snipd podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share & Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode