The Prompt: Making work more satisfying with generative AI

Business leaders are buzzing about generative AI. To help you keep up with this fast-moving, transformative topic, each week in “The Prompt,” we’ll bring you observations from our work with customers and partners, as well as the newest AI happenings at Google. In this edition, Philip Moyer, Global VP, AI & Business Solutions at Google Cloud, discusses how generative AI is reshaping today’s most valuable skills and changing how people do their jobs. 

It’s not news that disruptive technologies can change jobs and how we do them. But as generative AI becomes more adept at tasks that normally require a highly-skilled human, from creating code to passing standardized tests like the U.S. Medical Licence Exam (USMLE), executives are curious just how monumental incoming shifts might be. 

For example, in an analysis of over 900 occupations, Goldman Sachs economists estimated that roughly two-thirds of jobs in the U.S. could be impacted by AI. That’s an eye-popping number—but it’s important when looking at this kind of data to remember that “impacted by AI” isn’t synonymous with “replaced by AI.” 

Consider a recent report from the National Bureau of Economic Research (NBER) that found customer support agents became 14% more productive on average when they used a generative AI conversational assistant, with novice workers showing the greatest improvement. As data like this suggests, AI is often a bigger opportunity to extend human ability than to automate tasks. 

With that in mind, in this edition of “The Prompt,” I want to explore some of the ways generative AI is making the workplace happier and more fulfilling.

Make work more enjoyable by reducing drudgery and turnover

Many jobs are filled with rote, repetitive tasks like processing insurance claims, extracting pieces of information from a spreadsheet, or answering the same customer service questions again and again. Such tasks aren’t generally joyful and tend to eat up time employees would rather spend on more creative or challenging work. 

By automating or accelerating the tasks that people find most frustrating, generative AI has the power to make jobs more fun. Instead of spending hours crafting the same responses in a support center or combing through an inbox for important updates, workers can use generative AI to remove the toil and drudgery. 

AI assistants can automatically generate responses to common queries or help employees more quickly draft, summarize, and prioritize emails, for example. They can help ensure work complies with various regulations or make it easier to repetitively assemble large bodies of text from multiple sources, such as for requests for proposals (RFPs). These efficiencies can add up, helping workers to focus on more interesting work without constantly toggling through administrative tasks or wasting hours digging through emails for the right information. 

Such use cases can be particularly impactful for jobs that are extremely hard to hire for and retrain. If a role is difficult to fill or subject to frequent churn, that’s a good indication the job is not rewarding. That said, the NBER paper showed customer support agents were less likely to leave roles when given access to AI assistants,  indicating that by reducing repetitive tasks, AI can help employees to focus on more fulfilling work.

Increase human productivity to drive ROI

By amplifying human talent or acting as a catalyst for human productivity, generative AI can help drive revenue in areas where organizations are limited by resources or ability. 

People with great ideas don’t always have the time or experience to produce great messaging, for instance. But a generative AI assistant can quickly and easily transform brainstormed bullet points into blogs, one-pagers, or presentations. This ability doesn’t eliminate the need for skilled writers or editors, but it can significantly accelerate workflows by producing better drafts earlier in the ideation process and enabling communications professionals to focus on messaging instead of proofreading. 

Another great example is generating code via natural language prompts. This won’t diminish the value of talented engineers, but it can help them to work faster, and it can mean simpler apps that used to end up in IT backlogs can be prototyped without developer assistance. Across a range of tasks, generative AI means that instead of being slowed down by laborious processes, workers can dedicate time to the most valuable tasks and stretch their abilities in new directions. 

Maximize the value (and happiness) of your top talent

Building from the previous point, organizations should also explore how generative AI can eliminate scenarios in which highly-paid employees are blocked from creating value. Doctors spend hours converting clinical notes into something a patient or colleague can easily read and understand. Business executives regularly have to draft investment memorandums to attract and secure investment. Developers dedicate a large portion of their work to debugging and reviewing code, reporting, or helping others understand projects. 

While these tasks might be necessary, they take up an enormous amount of time, don’t maximize the abilities of highly-skilled workers, and certainly don’t add any delight to the day. Generative AI can help by accelerating repetitive work and empowering people to focus on tasks that demand their skills. 

Invest in human-centric AI

Though generative AI’s full potential is still revealing itself, the most powerful use cases will continue to put humans at the center. We’ll have more to say on this topic in coming months, including looking at some of the new jobs that generative AI is creating, such as prompt engineers, AI auditors, or machine managers. But the salient point is, if leaders focus on the delta between what employees like to do and what they often have to do, they can target generative AI use cases that make workers happier and more productive, and that unlock new value for the organization. 

This week in AI at Google 

• Helping robots navigate. Teaching robots to navigate complex outdoor environments is crucial to real-world applications, such as delivery or search and rescue—but it is also incredibly complicated because the robot needs to perceive its surroundings, then explore to identify possible paths towards a goal. Check out this blog post to learn how Google researchers are tackling this challenge with a learning-based transfer algorithm that uses deep reinforcement learning to train a navigation policy in simulated indoor environments and transfer it to real-world outdoor environments. 

• Get ready to NeRF out. Text-to-image is one of the most exciting generative AI use cases— but how about text-to-3D? Take a look at DreamFusion, a system built by optimizing a NeRF from scratch using Google’s text-to-image diffusion model. 

• Breathe easier with AI. Google research has previously explored how ML models can help detect retinal diseases—but these models were trained using labels from clinicians, which require time, expense, and expertise to create. In Inference of chronic obstructive pulmonary disease with deep learning on raw spirograms identifies new genetic loci and improves risk models, published in Nature Genetics, Google researchers describe a method for training accurate ML models even when using noisy and unreliable labels, including illustrating with ML models that can characterize lung function. Read this blog for the full scoop. 

• GitLab gets generative. GitLab, a major provider of enterprise DevSecOps capabilities, is leveraging Google Cloud’s infrastructure and generative AI foundation models to provide customers with AI-assisted features directly within its platform—see here for the details.