• Report claims 95% of companies have nothing to show for their AI investments
  • Only 5% have successfully deployed AI tools at scale
  • Workers prefer ChatGPT to custom AI

New data from MIT’s NANDA (Networked Agents and Decentralized AI) initiative has claimed that although US companies have invested $35-40 billion in generative AI tools, an overwhelming majority (95%) of them have nothing to show for it.

This leaves only 5% of companies having successfully deployed AI tools at scale, with failures blamed on AI’s inability to retain data, adapt and learn over time – not the shortage of infrastructure and talent that often reaches the headlines.

Exploring a range of deployments, from off-the-shelf to specially-designed systems, MIT found only 5% of custom AI tools ever reach production.

Companies don’t have much to show for their AI investments

With many execs now seeing demos as little more than science projects, confidence in AI initiatives has declined among corporate leaders.

The smallest impacts were measured across professional services, healthcare and pharmaceuticals, consumer and retail, financial services, and energy and materials.

Although many companies are struggling to quantify the benefits of their AI deployments, 80% of execs across tech and media are expecting reduced hiring over the next 24 months.

However, workforce impacts vary, with job cuts mostly affecting non-core and outsourced roles – an estimated 5-20% of such roles already impacted.

The study also revealed workers prefer generic tools such as ChatGPT over specialized offerings, even when they’re powered by the same models.

The familiarity and flexibility of ChatGPT in particular has driven shadow IT adoption, with companies urged to consider worker needs and adapt policies accordingly to increase security rather than ban them altogether.

On the flip side, enterprise tools are generally seen as more rigid and less effective, despite their typically higher costs.

Looking ahead, it’s clear that there’s value in a much simpler strategy. Rather than creating complex proprietary systems, tweaking widely available tools to adhere to company policies could offer much better ROIs while simultaneously reducing the amount of dedicated AI training workers might need.

You might also like