Businesses locked in expensive AI “arms race”

John Collins says there is no question that we are in an AI arms race.

He has worked in the IT industry for 35 years in various roles including software programmer, systems manager, and chief technology officer.

He currently serves as an industry analyst for research firm Gigaom.

The current arms race was spurred by the launch of ChatGPT in late 2022, Collins said.

Since then, many such generative AI systems have emerged, and millions of people use them every day to create artwork, text, or videos.

For business leaders, the stakes are high: generative AI systems are incredibly powerful tools that can digest more data in minutes than a human could process in several lifetimes.

Corporate leaders suddenly realized what they and their competitors could achieve with AI, Collins explained.

“Fear and greed are driving it,” he says. “And that creates an avalanche of momentum.”

With the right training, customized AI systems can help companies stay ahead of their competitors, whether through breakthroughs in research or by cutting costs by automating tasks currently performed by humans. may be possible.

In the pharmaceutical industry, companies are customizing AI to help discover new compounds to treat diseases. But this is an expensive process.

“We need data scientists and model engineers,” Collins explains.

These scientists and engineers need to have at least some understanding of the pharmaceutical areas that AI addresses.

And that's not all. “We need infrastructure engineers who can build AI platforms,” he continues.

Finding such highly skilled workers is not easy.

“There just aren't enough people who understand how to build these systems, how to actually make them work, and how to solve some of the challenges ahead,” said Andrew Rogoiski, director of innovation at the Surrey Institute for Human-Centered AI. “No,” he says. at the University of Surrey.

These challenges are so important that the salaries of those who can tackle them have reached “ridiculous” levels, he added.

“If we had the ability, we could produce hundreds of AI PhDs because people would give them jobs.”

In addition to skills shortages, simply accessing the physical infrastructure needed for large-scale AI can be difficult.

The computer systems needed to run AI for cancer drug research typically require 2 to 3,000 modern computer chips.

The cost of such computer hardware alone can easily exceed $60 million (£48 million), not including the cost of other necessities such as data storage and networking.

One of the problems for businesses is that this type of AI has emerged fairly suddenly. Previous technologies, such as the advent of the Internet, were built more slowly.

Large banks, pharmaceutical companies, and manufacturers may have the resources to purchase the technology needed to take advantage of the latest AI, but what about smaller businesses?

Italian startup Restworld is a recruitment website for catering staff, with a database of 100,000 employees.

Chief Technology Officer Edoardo Conte was keen on whether AI could benefit the business.

The company considered building an AI-powered chatbot to communicate with users of the service.

But Conte said that with thousands of users, “the costs become very high.”

Instead, we focused on a narrower issue: that candidates are not always able to express their experience in the best way.

For example, a candidate may not list waiter as a skill. But Conte's algorithm makes it easier to uncover additional information, such as whether they have previously applied for and won a standby role.

“AI can infer that they might be a waiter or might be interested in other waiter jobs,” he says.

One of the hurdles in hospitality recruitment is getting candidates through the interview stage.

Conte's next challenge is to use AI to automate and customize the candidate interview process.

The AI ​​may even have a “conversation” with the candidate and create a summary to pass on to the recruiter.

The entire process, which currently takes several days, could be sped up, during which waiters and chefs could find other jobs.

In the meantime, big companies will continue to pour money into AI projects, even if it's not always clear what they can accomplish.

As Rogoiski puts it, AI implementation is in a “Darwinian experimental phase,” and it's hard to know what the outcome will be.

“That's where it gets interesting. But I feel a little bit like we have to choose it,” he said, adding, “I don't know if we have a choice.”