In its latest report titled "Big Ideas 2024," Cathie Wood-led Ark Invest predicted that the convergence of hardware and software has the potential to reduce artificial intelligence training costs by 75% annually until 2030.
What Happened: The report was shared on Wednesday and touched upon multiple topics, including technological convergence, AI, Bitcoin allocation, digital consumers, and more.
While providing information about AI, the report stated that AI models like OpenAI's GPT-4, which perform incredibly well on various tasks, are expected to boost productivity significantly.
"Jolted by ChatGPT's ‘iPhone’-like moment, enterprises are scrambling to harness the potential of artificial intelligence (AI)," the report stated, adding, "AI promises more than efficiency gains, thanks to rapidly falling costs and open-source models. If knowledge worker productivity were to quadruple by 2030, as we believe is likely, growth in real GDP could accelerate and break records during the next five to ten years."
See Also: OpenAI’s GPT-4 Demonstrates Limited Utility In Bioweapon Development: Study
Providing more details about the potential reductions in AI training costs, the report cited Wright's Law, stating, " improvements in accelerated compute hardware should reduce AI-relative compute unit (RCU) production costs by 53% per year, while algorithmic model enhancements could lower training costs further by 47% per year."
"In other words, the convergence of hardware and software could drive AI training costs down by 75% at an annual rate through 2030."
![](https://editorial-assets.benzinga.com/wp-content/uploads/2024/02/01032652/Screenshot-2024-02-01-at-12.56.39%E2%80%AFPM-1024x434.png)
Why It's Important: Large language models like GPT-4, PaLM, Claude, Bard, LaMDA, and LLaMA continue flourishing, but creating and training these models is costly. It involves using thousands of Graphics Processing Units or GPUs to handle the massive amounts of data they learn from.
Just the GPUs alone can cost millions of dollars. For instance, OpenAI’s GPT-3 language model needed at least $5 million worth of GPUs for each training session, reported Forbes.
When asked about the training cost of foundation models at an MIT event in July, OpenAI’s co-founder Sam Altman mentioned that it was “more than” $100 million.
The forecasted drop in AI training costs by Ark Invest has the potential to make AI technologies more accessible to a broader range of industries, fuel innovation in AI, help businesses deploy more AI solutions to enhance efficiency and automate tasks, impact the current job market and increase integration of AI into our daily lives.
Moreover, it could also heighten focus on ethical considerations and regulatory frameworks.
Check out more of Benzinga’s Consumer Tech coverage by following this link.
© 2024 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
Comments
Trade confidently with insights and alerts from analyst ratings, free reports and breaking news that affects the stocks you care about.