AI integration and MLOps

There’s no doubt that the cost-cutting appeal of artificial intelligence is disrupting the business of outsourcing, with its main appeal over the years being cost reduction.  Now large outsourcing operations like Sourcefit, based in the Philippines with 1,300 employees, are integrating AI, as the technology eliminates some types of jobs and creates new ones too, according to Sourcefit CEO Andy Schachtel.

The hope is that outsourcing and AI will turn into a game-changing combination that marries traditional low-cost outsourcing expertise with AI’s advanced data analysis, process automation and decision making capabilities. Indeed, outsourcing firms themselves may hand over low value-added, volume-driven digital services to supervised AIs in order to focus on more complex and creative tasks.

“Certainly AI will have a disruptive effect on the outsourcing industry,” says Schachtel, “but it will continue to be an essential tool for businesses to gain access to talent at a reasonable cost.”

While “it’s certainly humbling to think of the careers that will be permanently impacted.” says Schachtel, it’s clear “certain roles will be diminished based on what AI can and cannot do well.” Among the new job positions Schachtel sees developing are engineers for marketing content creation, AI CX specialists to train models on specific customer experience projects, and AI systems specialists to design and administer AI systems for a wide range of processes across virtually every industry.

AI training, however, is emerging as an outsourcing pitfall. A recent study by the Swiss Federal Institute of Technology found that between 33 and 46 percent of gig workers paid to train AI models were outsourcing their work to AI, thereby embedding errors in old models into new ones. The low pay of these gig workers is cited as an incentive to taking shortcuts.

“The old axiom of ‘bad data in, bad data out’ certainly holds true with AI,” says Schachtel. “We must always be cognizant of how AI models are trained and what type of misinformation and biases may have been baked in. There is no substitute for comprehensive testing whether you’re training a model within a traditional machine-learning framework or if you’re working with a large language mode. QC is vital to any project of this nature but it can be successfully implemented with the proper processes, the proper tools and the right culture.”

Schachtel notes that AI outsourcing to a global company like Sourcefit comes with certain advantages. While location doesn’t matter in terms of being able to deliver services seamlessly from a technical standpoint, it does matter as far as the type of talent available in different countries.

“Our centers in the Philippines are particularly strong in customer service and design, whereas our East European office is stronger in application development, and our Dominican Republic office fills a strategic need for Spanish and French speakers,” says Schachtel, who is fluent in French and Japanese. Schachtel  founded the company in 2009 after spearheading the development of an international mobile-first social media network for Vodaphone he outsourced to the Philippines.

“When a company is considering outsourcing, it is important for them to research the type of skills that are most readily available in a particular country or region,” explains Schachtel. “Outsourcing AI processes to a global company like ours is definitely more cost effective than trying to do it in-house or outsourcing locally.”

The more things change, the more they remain the same. At the end of the day, outsourcing will still be about delivering quality services at a compelling price point.