Just like Apple and Duolingo, Google is now choosing to sever ties with third-party contractor Appen instead of taking the route of layoffs. This shift stems from the preference of artificial intelligence model developers for specialized generative AI services. It also signifies a broader trend of tightening the belt on AI training data expenditures.

Appen used to be crucial in training some of the world's top AI systems for Adobe, Salesforce, and Nvidia, but its fate took a downturn after a leadership change. The departure of CEO Mark Brayan, who had been at the helm of the company for seven years, marked a turning point. Power was transferred to technology veteran Armughan Ahmad, who has 25 years of experience, but this transition led to a series of executive departures.

Despite having a global platform with over one million freelancers in 170 countries, Appen saw a 13% decline in revenue in 2022. The company is striving to adapt to the emerging era of generative AI, citing "challenging external operations and macro conditions" as reasons for its decline.

Former employees, choosing to remain anonymous due to fear of retaliation, shared their insights with CNBC, attributing Appen's current challenges to years of lax quality control and a disconnected organizational structure. As generative AI gains prominence in the tech industry, Appen and similar companies find themselves at a crossroads, grappling with the consequences of their AI shortcomings.

Work Automation

In the early days of AI models, the process of collecting artificial data fueled the growth of contractors like Appen. However, with the rise of large language models (LLMs), the landscape has changed dynamically. These human-like AI systems, such as Google's Bard and OpenAI's ChatGPT, can now search for data from digital infrastructures to answer queries and even generate images as responses.

In a recent turning point, these language model companies have started reducing their spending on outsourcing work through Appen and similar companies. Instead, they are focusing on resetting their businesses through automation services provided by specialized generative AI companies like Labelbox and Scale AI.

Just four months ago, Google Cloud decided to extend its partnership with Labelbox, a platform that launches a suite of tools tailored for AI technology, such as reinforcement learning for AI feedback (RLHF), reinforcement learning for AI feedback (RLAIF), evaluation, and red teaming.

Meanwhile, another company in this field, Scale AI, has been solidifying its position among clients that include top-tier companies like Toyota, Instacart, PayPal, and Flexport. Scale AI has been making waves by offering customized data solutions and navigating the complex waters of the machine learning lifecycle.

The competition in this field is fierce and will only become more challenging from now on.

Excuses, Excuses

Google's shift from Appen to Labelbox is a classic example of why companies should quickly and decisively integrate generative AI and evolve with changing environments. The demand for AI-driven automation services is on the rise because they are cheaper, faster, and involve fewer issues.

Kim Stagg, Vice President of Product at Appen, publicly announced that the work required for generative AI services is different from what the company previously needed. It seems that the company failed to get on board the generative AI ship and was left behind.

While Appen's executives blame AI, former employees believe the problem lies in the organization's structure and the reshuffling of executives. This pattern has emerged in several companies, using generative AI as an excuse to cover up mishandling of internal affairs.

Economist Cory Stahle clarifies that AI tools are not yet sufficient to fully replace workers. Stahle states that if AI were truly to blame, layoffs would be more widespread across industries. "We haven't seen that happen yet," he says.