The world of Artificial Intelligence is moving incredibly fast. OpenAI’s recently developed Chat GPT is still taking the internet by storm. As a Natural Language Processing Model, it’s an artificial intelligence that can be asked unstructured questions and receive intelligent responses on a broad range of topics. Chat GPT is currently in it’s 3rd generation with a 4th generation GPT-4 coming soon to general availability and rumored to be powering the New Bing Search experience.
Adept.AI recently announced ACT-1 a General Artificial Intelligence model designed to interact with computers like a human would without manually building out scripts like you would in traditional Robotic Process Automation (RPA) methods.
The architecture behind both technologies are similar. Adept.AI utilizes a neural network, machine learning technique called a transformer. Transformers are much more efficient than prior machine learning models, and enable better performance. It’s the key technology behind many advanced Natural Language Processing bots.
You can Sign up to Join the Waitlist as soon as the technology is available to the general public.
Is Adept.AI the Chat GPT
The world of Artificial Intelligence is moving incredibly fast. OpenAI’s recently developed Chat GPT is still taking the internet by storm. As a Natural Language Processing Model, it’s an artificial intelligence that can be asked unstructured questions and receive intelligent responses on a broad range of topics. Chat GPT is currently in it’s 3rd generation with a 4th generation GPT-4 coming soon to general availability and rumored to be powering the New Bing Search experience.
Adept.AI recently announced ACT-1 a General Artificial Intelligence model designed to interact with computers like a human would without manually building out scripts like you would in traditional Robotic Process Automation (RPA) methods.
The architecture behind both technologies are similar. Adept.AI utilizes a neural network, machine learning technique called a transformer. Transformers are much more efficient than prior machine learning models and enable better performance. It’s the key technology behind many advanced Natural Language Processing bots.
Is Adept.AI the Chat GPT of Digital Process Automation?
Adept.AI recently published several videos to their blog showing natural language text input describing several common tasks on a computer. The AI recognized what the user was requesting and proceeds to complete the task without human intervention.
Some of the tasks shown are trivial if you’re a professional Microsoft Excel user, but for the many people who avoid that little green button, this technology could be a life saver. The technology also appears to be a big time saver, automatically drafting e-mails on command and completing many other tasks.
Take a look at their blog showing many additional videos of different use cases.
Robotic Process Automation (RPA) Vs. Artificial Intelligence (AI)
To understand how special Adpet.AI is, you have to understand the differences between RPA and AI. The two are often confused. There is some overlap which leads to even greater confusion such as an example where an RPA bot sends information to be processed using AI technology. But the two are wildly different technologies.
How Robotic Process Automation Really Works
The giant RPA software sellers want you to think that the technology is cutting edge and make you think that they’re using some kind of advanced technology to automate repetitive tasks. You will see terms like Software Robots which make you think of very high-tech processes.
In reality, RPA is a script that runs on a computer designed to automate high volume repetitive work. Scripts are typically created through a manual process where the designer goes step by step through a process and tells a computer what to do. They setup the script to interact with a computer in the same way that a human would.
Each script is designed to complete a specific task. If there is ambiguity or errors in the process, the bot will have to be programmed with decision trees or other rule-based criteria of how to handle exceptions. If the bot fails, it will require human intervention.
Because of the amount of programming required, RPA is mostly suitable for high volume tasks that are easily repeatable with few exceptions and minimal decision points.
Our RPA vs AI article goes into more detail about the differences and overlap between the two technologies:
What is the difference between RPA and AI? (www.popautomation.com)
Why Artificial Intelligence is a Great Use Case for RPA
It takes a long time to build out automation scripts. Even the best platforms require some computer programming knowledge, and the entire field can be incredibly expensive to get into. The biggest RPA providers can easily cost companies hundreds of thousands of dollars to get started, even before factoring in ongoing labor to support RPA initiatives.
A solution like Adept.AI instantly makes RPA approachable for all companies. A non-technical employee may not be able to program an RPA Bot, but they can usually describe what they’re trying to accomplish on a computer. If an AI is able to complete tasks on command, it will be a massive efficiency gain for nearly every company that currently works with computers.
How does Natural Language Processing (NLP) Work?
The introduction of Chat GPT to the public was a game changer. The results that it’s able to accomplish as a language model based on machine learning techniques has been incredibly impressive.
Chat GPT was fed a massive amount of sample data. A machine learning model takes sample data and splits in into two parts. One part has the input data, and the desired output data. The other part has only input data, and the computer algorithm goes through iteration after iteration until it figures finds an optimal algorithm that bridges samples data into the desired output.
If an incorrect output is given, then a human gives the machine feedback and trains it.
What are the Challenges for AI Driven RPA?
Using machine learning systems like Chat GPT that take text data as an input and pass it through an output layer to generate text is an immensely difficult task.
For an AI driven RPA solution to work, Natural Language Processing is only the beginning. To be truly useful, an AI would have to accurately interpret text input, combine it with image recognition to determine what is happening on the screen, and have enough training data for each specific piece of software that it would know what buttons to press or click.
While there is a lot of information on the internet and other sources currently available to help Chat GPT work with Power BI and guide human users through various tasks, we don’t know if there would be enough training data for most computer programs for the system to become truly useful in the near future.
What is the Future of Adept.AI and AI Automation Systems?
We are still in the early days of the Artificial Intelligence Revolution that’s coming to businesses across the globe. It’s like that Adept.AI will continue their ongoing work, or competitors will surface in the near future that will inch us ever closer to re-creating the usefulness of the human brain.
Companies like Tesla have been working on autonomous vehicles for close to a decade. They still have a lot of flaws, and nobody knows how close they are to becoming a reality. However, if you asked us ten years ago if a self-driving car would ever make it to the real world, we would have said no. Companies like Waymo and Cruise have driven over a million miles each. So we wouldn’t count out AI powered RPA coming our way in the not-so-distant future.
Tech Crunch did a deep dive into the background of the startup, if you’d like to learn more and follow this exciting new technology:
Adept aims to build AI that can automate any software process | TechCrunch