The latest Stanford AI Index suggests the cost of training advanced AI models has surged, generative AI investment has skyrocketed and industry continues to dominate AI development.
AI continues to be the big buzzword in the tech sector, with industry focus – and industry costs – growing as a result.
That’s according to the AI Index – an independent initiative at Stanford University – which has compiled its annual report on global AI developments. This is the initiative’s seventh edition of this index and includes a broader overview of the sector – as “AI’s influence on society has never been more pronounced”.
The report claims that global private investment in AI decreased for its second consecutive year, but that investment into generative AI “skyrocketed” – an unsurprising discovery given the wave of generative AI products and hype witnessed worldwide since the rise of ChatGPT towards the end of 2022.
Meanwhile, the growth of massive “state-of-the-art” AI models has also led to increased costs to train them, according to the report. The AI Index estimates that these costs have reached “unprecedented levels.
“For example, OpenAI’s GPT-4 used an estimated $78m worth of compute to train, while Google’s Gemini Ultra cost $191m million for compute,” the report said.
Both of these models were released last year and represent a substantial leap in costs compared to the previous cost leader – Google’s PaLM model – which cost more than $12m worth of compute to train in 2022.
Industry continues to dominate
Last year’s AI Index noted the dramatic shift AI development took in the previous decade, as industry began to heavily dominate the sector while academia fell behind – a reversal of the state of AI development in 2014.
The latest AI Index claims industry continues to dominate “frontier AI research”, as industry members produced 51 notable machine learning models in 2023 compared to only 15 produced by academia.
The report highlighted growing links between academia and industry in this sector however, as collaborations between the two groups created 21 notable machine learning models last year – a new annual record according to the Index.
Meanwhile, the number of foundation models being created worldwide is growing rapidly. The report claims 149 foundation models were released last year, which is more than double the amount released in 2022.
“Of these newly released models, 65.7pc were open-source, compared to only 44.4pc in 2022 and 33.3pc in 2021,” the AI Index said.
Concerns around AI
While various industries have been quick to adopt AI into their products and services, there have been growing concerns around the dangers this technology poses. The AI Index noted that new regulation has passed around the world that focuses on AI, particularly in the US.
But the report also claims more people believe AI will dramatically affect their lives in the next three to five years – and a substantial number are nervous about AI products and services.
A survey by Ipsos suggests most of the public are pessimistic about the economic impact of AI. The AI Index also claims that AI sentiment in Western countries is low but is “slowly improving”.
“In 2022, several developed Western nations, including Germany, the Netherlands, Australia, Belgium, Canada and the US were among the least positive about AI products and services,” the report said.
“Since then, each of these countries has seen a rise in the proportion of respondents acknowledging the benefits of AI, with the Netherlands experiencing the most significant shift.”
Meanwhile, the AI Index claims robust evaluations for large language models are “seriously lacking” and that there is a lack standardisation in responsible AI reporting.
“Leading developers, including OpenAI, Google, and Anthropic, primarily test their models against different responsible AI benchmarks,” the report said. “This practice complicates efforts to systematically compare the risks and limitations of top AI models.”
Find out how emerging tech trends are transforming tomorrow with our new podcast, Future Human: The Series. Listen now on Spotify, on Apple or wherever you get your podcasts.
