PayPal senior VP: We’re now in the ‘intelligence era’ and companies should be focused on tokens

“We’ve moved on [an] “From the information age to the intelligence age,” Prakhar Mehrotra, PayPal’s senior vice president and global head of artificial intelligence, said at the Fortune Brainstorm AI conference earlier this month.

This “age of intelligence” is characterized by industries moving away from the data storage and retrieval model, Mehrotra said. luck Journalist Sharon Goldman. Instead, because of AI capabilities, data can be generated more spontaneously, with the ultimate goal of achieving autonomy in some parts of the workplace.

Companies are racing to implement artificial intelligence — with its promise of increased productivity and output — in their workplaces, but their successes have been mixed. An MIT study in August found that 95% of workplace AI initiatives in organizations failed to deliver rapid revenue acceleration.

“It will be a journey… you have to go through this crawling, walking and running,” Mehrotra said. “I think that adage was true 10 years ago, and it’s also true in this day and age.”

The future of artificial intelligence factories

The future of building AI in the workplace will be investing in AI factories, on-premises or in the cloud, said Mark Hamilton, vice president of solutions architecture and engineering at Nvidia who was interviewed alongside Mehrotra at the conference. This is because the data needed to run companies will not be primarily retrieved by humans or computers, but rather generated by artificial intelligence.

“When you go in and say, ‘Make a PowerPoint slide that says this,’ or ‘I’m working on this coding job, can you go in and create the code?’ It’s not extracting it from the database, it’s taking a model and generating that data,” Hamilton said.

Mehrotra noted that for companies to effectively build the computational power needed to generate this data, there must be a new atomic unit that companies value: tokens, or the basic component of text that AI needs to understand and process language. Tokens are a snippet of information used to train the data, as well as what is generated by the AI ​​after the model receives a prompt.

“Every company should think about their data in terms of codes, because then… [they] “You can derive that intelligence from it,” Mehrotra said.

Token generation, a measure of inputs and outputs, has become a key metric for technology companies in particular. In May, Nvidia boasted that Microsoft, which uses Nvidia chips, produced more than 100 trillion tokens in the first quarter of this year, a five-fold increase year over year. These production indicators could help AI companies sell themselves to investors and boost valuations, although data shows the tokens’ correlation with demand and profits is weaker than the tech companies suggest.

Mehrotra and Hamilton agreed that many companies today see the value of tokens in enhancing AI capabilities, but are examining how they can better fit their needs, such as which tokens should be acquired or purchased, what should be created within the company, and for what purpose? Next, each company has its own AI factory, where it receives tokens and outputs tokens of value.

“I see it as just building that muscle,” Mehrotra said. “Like if all employees start thinking in terms of codes, in terms of the generation process, then, yeah, it’s a different company.”