Companies find AI rollout is slower, costlier, and needs more people

In short: Many companies are still interested in AI, but rolling it out widely is slower and more expensive than early hype suggested, and it takes ongoing human work.

What's going on

Companies are not abandoning AI, but many are finding it hard to move from small tests to everyday use across the business. Several reports say projects often get stuck in the “pilot” stage, which is like a trial run that never becomes the normal way of working.

A big reason is that AI has to fit into existing software and data systems. Zapier reports that 78% of large companies are struggling to connect AI tools to what they already use. Aura Intelligence reports that over 90% of organizations have integration problems, and 74% struggle to get results that scale.

Cost is another barrier. Some leaders say the price of AI vendor tools is too high, and others say ongoing support is the bigger issue. AI can also require upgrades to infrastructure such as cloud computing, special chips called GPUs (hardware used to process large amounts of data quickly), and better data systems. It can feel like buying a powerful new appliance and then learning your home wiring needs an expensive upgrade.

Human effort is also a bigger part of the work than many expected. Companies need people who can prepare data, connect systems, and watch for mistakes and risks. That includes monitoring how AI performs over time, like a manager checking quality instead of setting a machine and walking away.

What to watch

Expect more companies to talk openly about the practical limits of AI and focus on smaller, clearer projects that save time or money. Also watch for more spending on tools that help manage and monitor AI in real workplaces, such as governance and oversight software.

Source: NYTimes