An introduction to how ACT-1 is working towards prioritizing the user before everything else.
You must install software on computing systems (such as macOS and Windows) to accomplish anything. Ocean of problems, ocean of solutions, and all the responsibility on the user — you must know where to start!
You must also understand how to locate a bunch of software and tooling, why you should use them, and how to use them.
Prior to interacting with a computing system, we are already confronted with a plethora of challenges. In order to begin addressing them (those problems), we must have or attain a certain level of experience across all of our computing systems.
Yet, at the same time, we remain highly anxious and eager for the simplest passageways, most straightforward routes we can take for answers.
ACT-1, also known as Action Transformer, evaluates problems based on large language models (LLM), surgically weaves few-shot and zero-shot learning procedures, and pushes all of that behind the curtain to construct a staged user experience (UX) for you.
Why is it necessary to load software before considering how to solve a problem?
Why must you first visit a large number of websites to acquire the information you require? How well do you know which websites to visit first, second, and to start?
When you attempt to talk to Siri, a verbal answer is provided (assuming this is how you configured it, but let’s leave the technical details aside).
Similarly, this type of question answering may be transformed into a physical object. Speak to your speaker that is ten feet away; it may respond in response to your question.
Other use cases include people physically interacting with technology to complete actions. Of course, such “tasks” necessitate your utilization of the aforementioned computing systems, which are also your preferred method of solving for your problem. In other words, you want to use your MacBook Pro.
Consider a use case in which you require assistance locating a multifamily home with at least 2 acres within a specific budget. To do such a search, you may modify the filters on a website specializing in property searches (and wait for the results to populate). What if, instead of performing all of these lengthy administrative steps up front, you could simply compose your preferred search query in a single sentence and let ACT-1 perform the other duties before displaying the results?
“Users’ expectations about discoverability… are set by their prior experiences with all the other software they have used.”
Discoverability refers to the simplicity with which users can locate the information they seek in your design. This involves the ease with which users may identify information, menus, controls, and connected content or products. If users are unable to discover what they need quickly and effectively, their experience with the product will be negative.
The placement of the microphone visualization, in addition to the statement regarding “Enter a task [sic],” point to straightforward actions you may consider. This is so uncomplicated that you may focus your concentration on your needs-informing ideas rather than dealing with hundreds of program navigation faults and problems.
Some high level opportunities for discoverability against such interface design may include the following:
— Use concise and clear labeling: Labels should be simple to comprehend and clearly reflect what the user will discover when clicking through.
— The navigation bar should be structured in a way that makes sense and is straightforward to use. Consider employing dropdown menus to organize similar things into bigger categories.
— Use visual cues to direct the user: Icons, colors, and other visuals can help highlight essential objects and connections.
— Include search functionality: A search bar is essential for assisting users in quickly and easily locating the information they want.
— Implement filtering options: Filtering tools enable users to efficiently narrow down their selections so they do not feel overwhelmed by too many possibilities.
“The notion of affordances implies that the user is matching his or her goal against the set of opportunities offered by the environment, that he or she directly sees what can be done to reach the goal.”
The computing system must be clear and obvious to grasp so that users can rapidly learn how it works and utilize it efficiently. For instance, a hamburger icon may make it clear that clicking on it would send you directly to that hamburger’s selection page (this would be an example of high-level functionality offered by UX without having special instructions from users.)
Conversely, bad UI may force users to hover over an option in order for it to become apparent and evident, making the activity harder and more time-consuming.
Affordances are crucial in user interface design because they may make it easier for users to grasp how a system works without specific instructions and improve the overall usability of a design by making activities more clear and less time-consuming.
Simply by glancing at the interface Adept built, I can tell that the two alternatives accessible to me, speaking and typing, do not actually necessitate that I get a handbook with descriptions on what they signify. A door knob should open the door (particularly open it depending on its physical design); such an analogy relates to the design of ACT-1.
There is no necessity that artificial intelligence have a physical form or humanoid appearance; nonetheless, what would prevent ACT-1 from acquiring a physical form in the future (and what will become those corresponding use cases)?
If you have any recommendations for this post or suggestions for broadening the subject, I would appreciate hearing from you. Simply, send me a private message anywhere on this post.
Also, here is my newsletter; I hope you will kindly consider subscribing.
I have authored the following posts that you might find interesting:
References: this post contains text-embedded citations.
Read the full article here