Previously, Meta aimed to spearhead organizational transformation by leveraging AI, with a vision to streamline product development teams to just 3-5 members—a move that would have seen some teams face layoffs of up to 60%. Concurrently, the company embarked on an ambitious project to train AI Agents by meticulously recording employee operations, with the ultimate goal of automating human tasks. However, post-implementation, the reality fell far short of expectations. The volume of ineffective code modifications skyrocketed, while tangible improvements in user-facing functionalities remained scarce. Moreover, the frequency of technological safety incidents spiked, leading to a significant increase in employees' troubleshooting times.

In light of these challenges, Meta made the strategic decision to suspend its second round of layoffs and restructuring, originally slated for November. The company also scaled back its AI training initiatives and introduced welfare measures to bolster employee morale. Similarly, Microsoft, after a large-scale (or mass-scale) promotion of AI tools, encountered setbacks. The anticipated business outcomes failed to materialize, with issues such as AI-generated content looping and a lack of originality in output. In response, Microsoft adopted a more measured approach, meticulously tracking AI expenditures, drastically slashing spending estimates, and imposing stricter quotas on employees' AI model usage.

Today, as a growing number of companies rush to deploy AI, a host of problems have surfaced. Many have experienced AI-related incidents that have had a substantial negative impact. Numerous practitioners now acknowledge that if they were to redeploy AI, they would adopt a more cautious and incremental strategy, recognizing the importance of balancing innovation with practicality.