"Digital labor" used to be a synonym for the gig economy. You were part of that movement if your work was done online, or if your jobs were assigned by a central platform in the cloud. I've been living that life for nearly two decades now; popular examples of this digital labor approach include the Uber (UBER -1.88%) ride-sharing service and the Fiverr International (FVRR -4.19%) gig-connecting service for digital freelancers.

But the term found a new life in the last two years. When the ChatGPT artificial intelligence (AI) platform came along, some would call it "digital labor" when a human crafted effective prompts to be fed into generative AI systems. And now the AI revolution is taking the next logical step -- in 2025, digital labor often refers to automated AI agents that can set their own tasks and plan out complex data management activities.

This form of digital labor can help people achieve more with less work. Critics argue that it's a slippery slope, perhaps leading up to self-aware AI systems with dangerous priorities. Is digital labor the first step into a better future, or is the Terminator movie franchise's SkyNet coming up next?

Why ChatGPT won't become Skynet anytime soon

First of all, the AI industry is a long way away from self-aware machines. ChatGPT may be good at summarizing long texts and its sister service Dall-E 3 usually puts the right number of fingers on people in its generated images, but that's as far as it goes.

  • Even the best image generators often get basic details wrong in 2025.
  • I've seen a ton of AI-generated text, made by the best large language models (LLMs) with expertly crafted prompts and highly specific, high-quality data. The results are still repetitive, clumsy, and full of incorrect analysis.
  • Tech giants like Alphabet (GOOG -0.67%) (GOOGL -0.79%) and Tesla (TSLA 0.15%) have been working on self-driving cars for years. These systems get better all the time, but truly autonomous vehicles aren't a thing yet. Experts in the field think self-driving vehicles will be safe and useful by the mid-2030s.
  • I just started a chess game with the newest version of Alphabet's Gemini LLM. The AI's second move broke the rules of how knights move. Multi-function LLMs are not ready to play a proper game of chess yet.
  • The grand ambition of building artificial general intelligence (AGI) is also far away. OpenAI CEO Sam Altman currently expects computers to match human creativity in the 2030s, but experts have been overly optimistic about long-term targets before. Where's my flying car, you know?

Quantum computing could accelerate this progress, as non-binary calculations can find patterns in the noisy data of life much faster than old-school computers. But it will take a very powerful quantum computer to make meaningful contributions in areas like AGI, human-like creativity, and robots that can act like ordinary people. McKinsey researchers expect thousands of usable quantum computers by the year 2030, but truly complex problems will still be out of reach until 2035 or 2040.

So the world is about 10 years away from an avalanche of game-changing leaps in the AI and quantum computing realms. I'm looking forward to that sea change in human progress, and I can only hope that the new systems are developed with long-term security in mind. Isaac Asimov's three laws of robotics should be a daily reading in every college and high school until then.

But wait -- AI agents are on the move in 2025

But there will be progress, and it's happening right now.

AI hardware experts Nvidia (NVDA -0.02%) and Advanced Micro Devices (AMD -4.31%) have already developed next-generation AI accelerator chips that can run circles around the stuff OpenAI used for training ChatGPT 3 and 4. Every tech titan worth the name is working on better AI software and services.

Nvidia's financial results are soaring because a planet full of ambitious entrepreneurs is pushing the limits of the current technology, in search of whatever comes next.

Digital labor is an important piece of this puzzle. The more you automate your AI training and development efforts, the faster you'll see real progress. AI agents can help with that. Their history will always be a balancing act on the borderline of fast automation and quality results. Mistakes will be made, setbacks are coming -- and the overall progress will continue anyway.

Who's who in AI automation

If you have been following the ChatGPT-inspired boom in generative AI, many of the leading names should be familiar in the AI version of digital labor.

OpenAI is an early leader with AutoGPT, an LLM that can generate complex query plans to tackle difficult research issues. Alphabet has included an experimental form of the same approach in its latest Gemini models. UiPath (PATH 1.74%) has always sought to automate digital business processes with AI assistance -- AI agents just take this idea to the next level.

The list goes on, and will surely expand in 2025.

Now you know what digital labor and AI agents are all about. These ideas should be safe and helpful for at least the next decade, and very smart people will set up safeguards against potential threats (with AI agent help!) along the way. And you can invest in many of the key players right now.