A decade on, Silicon Valley is embarking on its biggest ever dogfooding experiment. This involves Big Tech companies taking large language models and generative AI tools and putting them to work inside their own organizations.

The implications are huge. These dogfooding tests will help decide how successful AI products really are in practice. Trillions of dollars are riding on this. The findings could also change how Big Tech companies operate — and how many expensive engineers they need. Millions of jobs are riding on this.

A golden AI 'Goose'?

Google President Ruth Porat hinted at this during a recent earnings call when she discussed "streamlining operations across Alphabet through the use of AI." 

Business Insider's Hugh Langley took the hint and found out about a new AI model Google has quietly rolled out internally. It's called Goose and it's designed to help Googlers write software code quicker and better.

Copilots inside Microsoft

Microsoft just embarked on a similar grand internal experiment. BI's Ashley Stewart reported recently that the software giant has been rolling out its new 365 Copilot AI productivity upgrade to employees.

One big team at Microsoft had a recent "hackathon" to get employees up to speed about the latest techniques in deploying AI, Stewart wrote.

A good way to convince customers

The promise of LLMs and generative AI is that software will become so good at automating tasks that companies will either churn out products way faster or make the same amount of stuff with fewer employees. Or both.

Big Tech companies want to sell AI tools to help businesses, developers, advertisers, creators, and other customers reach this new productivity nirvana. What better way to convince customers than showing them how you've already used AI to streamline your own operations?

A question mark over future hiring

If this AI technology lives up to the hype, we could see thousands of highly paid software coders and other knowledge workers lose their jobs. Or, current employees could keep their jobs and produce a lot more stuff.

Either way, there's now a big question mark over highly paid tech jobs. There was a job boom across the industry over the past 2 decades. The next 20 years is looking less rosy on that front.

If these dogfooding tests go well, Big Tech companies may not need to hire as many workers in the future.

For example, last year, Google infrastructure veteran Urs Hölzle said in an internal email that the company would use automation to "find more efficient ways of doing things." That included reducing the ratio of site-reliability engineers to software engineers.

There have been some heavy tech layoffs over the past year. However, Big Tech companies still have a lot of employees who get paid very well.

At the end of 2023, Google had 182,502 employees. That was down by less than 8,000 from a year earlier. Microsoft has more than 200,000 employees.