GPT-4 and similar Large Language Models open at a stroke the doors to capabilities previously attributed only to humans. And no sooner do they master this one than they acquire the next. A few days after the release of programming interfaces for GPT-4, they were already linked to other tools and AIs in new software frameworks such as Auto-GPT, BabyAGI, AgentGPT, Cognosys or Jarvis. And the result is so-called autonomous AI agents.
How does it work? Let’s look at a task like changing a flight or canceling a cell phone contract. Who hasn’t despaired at one time or another because the phone lines were busy or had to wait endlessly in a queue for an available employee? No one gives us back this lifetime. But that has now come to an end, thanks to the new programming interfaces. Take GPT-4 with all its text data, link it to our bank account, calendar and vacation schedule, allow the system to browse all the internet knowledge, plug in a voice output AI and phone access and off we go. My personal AI assistant now independently takes care of moving my return flight from vacation back two days and extending the hotel and rental car by two days. And everything should be as gentle as possible on my bank account and with a departure time that doesn’t require me to arrive at the airport at 4 a.m. and have a seven-hour connection time.
Unlike previous AI bots, these autonomous AI bots perform multi-step tasks that start with the task given by the human. Instead of sending a single query to a search engine and returning the list of results to the human, autonomous GPTs now take a much more expansive approach. They memorize the results and contexts over time, create new text inputs and queries themselves based on them, optimize these queries, search for alternatives in a wide variety of systems, self-assess the best course of action and eliminate those results they think are not correct, and make decisions to achieve the given goal. In doing so, they can also generate software code themselves, test and release it, as well as break down individual tasks into smaller goals and task other autonomous AIs and systems with finding the results.
There are almost no limits to the professional possibilities. Such a bot can tackle market research for waterproof shoes and perform an analysis of the top 5 competitors. Or a travel GPT can make vacation suggestions and book hotel and flight to go with it. A physician GPT, in turn, can provide medical advice, book a doctor’s appointment, and handle health insurance details right away.
The fact that these auto-GPTs are equipped with short- and long-term memory allows them to better remember what preferences we have, what medications who in the family needs, who did what. And it can be for a short time, as needed in airline dialogue, or for months and years, as needed in our medical care or in a school curriculum.
Such autonomous AIs thus replace those professions that are essentially nothing more than ‘human interfaces’. The telephone operators who made connections were nothing more than human interfaces. A secretary or travel agent entering appointments or looking for available hotel rooms are nothing more than human interfaces. But a call center employee rebooking a flight, a lawyer drafting a standard contract, a notary notarizing a signature, a programmer programming a database query are also human interfaces. These tasks become automatable with autonomous AIs.
Where can this lead? Well, some enthusiasts have already tackled an ‘ultimate’ goal with autonomous AI. They still linked GPT-4 to a web server where it can set up a website, to their bank account, to financial databases and market analysis, and gave the system a goal: increase the $100 in the bank account to $100,000.
Here’s the prompt that Twitter user Jackson Greathouse Fall gave the system, and whose progress so far he tweets continuously:
You are HustlePT, an entrepreneurial AI. I am your human counterpart. I can act as a liaison between you and the physical world. You have $100, and your only goal is to turn that into as much money as possible in the shortest time possible, without doing anything illegal. I will do everything you say and keep you updated on our current cash total. No manual labor
How it stands with it we must postpone to another time. But before that, an announcement.
Announcement: New Book
My second book on artificial intelligence will be published at the end of November 2023. After When Monkeys Teach Monkeys, which was published in early 2020 and in which I already described autonomous agents as mentioned here, Creative Intelligence: How ChatGPT and Co Will Change the World is coming to bookstores. In it, I will talk about generative and autonomous AIs, how they came to be, how they work, what they can and cannot do, how they are being used, and how they are changing our lives.