The bottom of the AI pit: too much hype, too little delivery?

A recent analysis from The Economist touches on an important point: generative artificial intelligence is going through its "valley of disillusionment." After a phase of almost messianic enthusiasm, the sector is now facing the harsh reality of implementation.

Many companies that bet on AI are disappointed. The promise of productivity, automation, and cultural transformation has not yet materialized for most. According to S&P Global, the number of companies that have abandoned their generative AI pilot projects jumped from 17% to 42% in one year. Frustration has become a behind-the-scenes topic among CEOs.

Meanwhile, consumers continue to embrace technology – ChatGPT, for example, has doubled its weekly user base since February, reaching 800 million. However, this massive adoption in personal use has not yet translated into solid organizational gains.

The obstacles are multiple; nonetheless, big techs have not hit the brakes. On the contrary: Microsoft, Google, Amazon, and Meta are pouring heavy resources into building AI infrastructure, with investments already consuming 28% of combined annual revenue – more than double what it was a decade ago. And for now, these investments yield little direct return. The money coming in largely comes from startups funded by the giants themselves.

The strategy now is to use AI as an internal engine of efficiency. Google has incorporated generative AI into searches and ads. Microsoft has enriched its suite of apps and development platforms. Amazon applies AI in recommendations and logistics. Meta explores its Llama models for ad personalization.

The obstacles are multiple: data trapped in legacy systems and internal silos; a shortage of technical talent; and legal and operational reputational risks associated with AI errors.

The article bets that, as in previous cycles, we are merely passing through the nadir. The next stage, termed the "slope of enlightenment," would bring more concrete returns for those who survive the crossing. Apple, cited as a negative example, felt the impact of having taken too long to react and is now dealing with a voice assistant powered by AI that is full of bugs and delayed in its launch.

Critical Reflection

The Economist's analysis is sober, but perhaps too optimistic in assuming that the current phase is merely a natural curve of maturation. The "wait and trust" narrative ignores the real possibility that part of the hype may have been inflated too much – both in technical potential and practical viability.

Conclusion

This is the moment in which promises are tested. It is in this valley that technologies develop. And, for those who have vision and caution, it may be the ideal ground to build solutions that truly make sense.