In the mid-20th century, the world entered the information age and industry transitioned to information technology. This era began with the miniaturization of computers and culminated with the invention of the World Wide Web, which put information at nearly everyone's fingertips. According to some technology leaders, the rise of AI marks the end of that era and the beginning of a new era of technology.
“We have transitioned from [an] From the Age of Information to the Age of Intelligence,” Prakal Mehrotra, senior vice president and global head of AI at PayPal, said at the Fortune Brainstorming AI conference earlier this month.
This “age of intelligence” is characterized by the industry moving away from a data storage and retrieval model, Mehrotra said. luck Reporter Sharon Goldman. Instead, thanks to the capabilities of AI, data will be generated more spontaneously, with the ultimate goal of achieving autonomy in some parts of the workplace.
Companies are racing to apply AI to their workplaces with the promise of improving productivity and output, with mixed results. An August MIT study found that 95% of companies' AI workplace initiatives fail to achieve rapid revenue acceleration.
“It's going to be a journey…we're going to have to crawl through this crawl and walk and run,” Mehrotra said. “This adage was true 10 years ago and it's true today.”
The future of AI factories
Mark Hamilton, NVIDIA's vice president of solution architecture and engineering, who was interviewed with Mehrotra at the conference, said the future of building AI in the workplace will involve investing in AI factories on company premises or in the cloud. The data needed to run a company will no longer be primarily obtained by humans or computers, but will be generated by AI.
“When you say, ‘Please generate a PowerPoint slide with this on it,’ or ‘I’m working on this coding feature,’ can you go in and generate the code for me?’ Instead of pulling data from a database, you take a model and generate that data,” Hamilton said.
Mehrotra pointed out that for companies to effectively build the computational power needed to create this data, they need a new atomic unit to focus on: tokens, the basic elements of text that AI needs to understand and process language. Tokens are both pieces of information used to train data and generated by the AI after the model receives a prompt.
“Every company needs to think about their data in terms of tokens. [they] You can extract that intelligence from there,” Mehrotra said.
Token generation, a measure of input and output, has become an important metric, especially for technology companies. Nvidia boasted in May that Microsoft, which uses Nvidia chips, generated more than 100 trillion tokens in the first quarter of this year, a five-fold increase year-over-year. These production metrics could help AI companies market themselves to investors and boost their valuations, but the data shows that the correlation between tokens and demand and profits is weaker than tech companies suggest.
Mehrotra and Hamilton agreed that many companies today see the value of tokens to improve AI capabilities, but are also considering which tokens to acquire or purchase, what to generate internally, and for what purposes to best fit the tokens to their needs. Therefore, every company has some kind of its own AI factory that both accepts tokens and outputs valuable tokens.
“I think it's just building muscles,” Mehrotra said. “If all your employees start thinking in terms of tokens, in terms of generation processes, then yes, it becomes a different company.”
