Should Skynet learn to code?
Should Skynet learn to code?
Issue 131: Will AI lead to nuclear holocaust or take our tech jobs?
Happy Sunday and NBA All-Star game for those who celebrate. This week I’m talking about Artificial Intelligence (AI) with the Chat GPT-3 craze at its peak conversation point. Open AI started as a non-profit
has now partnered with Microsoft, becoming the first existential threat to Google’s search business. Can you believe there is a world where people are downloading Microsoft Edge and using Bing on purpose? That’s what the AI hype is causing people to do! Let’s talk about the different applications of AI, how it will impact us as designers and builders, and why I’m not worried—viewing it as a collaboration instead of our new overlords.One of my fondest memories for me as a child was seeing the 1991 summer blockbuster “Terminator 2: Judgment Day” on the big screen. The first two films from the James Cameron sci-fi series are my favorite films of all time. In addition, I have a guilty guilty pleasure in the short-lived television series, "The Sarah Connor Chronicles." With the cancelation of Firefly, Serenity, and The Sarah Connor Chronicles, it turns out the true Terminator is Summer Glau
.If you’re not familiar with the acclaimed series, the first Terminator starts with a computer network named Skynet—an artificial intelligence created by the military (of course they did) that becomes self-aware and identifies humans as a threat (of course it did). On August 29, 1997, on Michael Jackson's birthday, Skynet launches a nuclear attack on Russia with anticipation that a counter-attack from Russia would wipe out the United States. The human survivors were rounded up as slaves, and forced to build machines for Skynet. Until one day in 2029, a man sparked resistance to fight back against the machines…John Connor. With the humans winning the war, the last hope for Skynet was to send a cybernetic organism with an Austrian accent back through time to assassinate John Connor’s mother Sarah. The rest of the films are basically more terminators and more Connors.
The trope of AI taking over the world and destroying humanity was nothing new before the Terminator. You might be familiar with HAL-9000 from 2001: A Space Odyssey, Star Trek: The Motion Picture, or Westworld (1973). So far, we seem safe from the nuclear holocaust and robot uprising, but the speed of AI’s acceleration has people concerned. Many people (myself included) believe that AI is going to transform how software is built. With the technical preview of GitHub Copilot in the summer of 2021, I began to ponder, “Should Skynet learn to code?”
Common uses of AI in software
There are heaps of uses in AI beyond what I cover such as robotics and autonomous vehicles. Here are the three uses of AI I believer will impact designers and builders.
Generative AI
This type of AI is used to generate new and original content, such as images, videos, and music, using techniques such as deep learning and neural networks. This is what you see with Midjourney and Dalle.
Natural language processing (NLP)
NLP is used to enable machines to understand and interpret human language, including speech recognition, language translation, sentiment analysis, and text generation. You might see this when using Chat-GP3 or suggested autocomplete while using Gmail.
Machine learning (ML)
ML involves the use of algorithms to analyze and learn from large datasets. It is used in various applications such as predictive modeling, fraud detection, and recommendation systems. You may not even be aware when machine learning is applied as it can feel invisible. If you use an iPhone and notice the smart search in your photos app and image recognition, that's machine learning in the background.
The paradigm shift
Paradigm shifts are nothing new in the field of art and design. During the Renaissance period, camera obscura became a common practice. In Leonardo Da Vinci’s journals, he writes about a pinhole projection technique that precisely describes camera obscura. For some, the technique was viewed as blasphemous. The old masters would never cheat like that to create art, would they? Like camera obscura, opaque projectors, and AI, these tools guide the intention of a creator’s outcome.
The technological shift
Model/View/Controller (MVC) is a common software pattern used for building interfaces, separating the data, user interface, and logic of how the application interacts. With AI MVC might look more like this:
Data sets (Model)
Contextual interfaces powered by AI (View)
Machine Learning algorithms (Controller)
The experience design shift
The practice of value-centered design sits at the center of the customer, technology, and business. If there was one area designers typically sit distant from (they shouldn’t), it’s technology since it’s often viewed as a developer thing. However, as AI and ML become naturalized in technology, it’ll be important for designers to increase their acumen in this space due to the implications it has. Paola Antonelli said in the documentary Objectified that she sees designers thinking like policymakers. This is the reason I encourage designers to embrace cutting-edge technology and have an open mind about it. It’s important for us to think about the ethical implications and systems design of such emerging topics.
Systematic and artisanal creation can co-exist
Galileo AI recently launched an impressive beta of their product—copilot for interface design. In this demo below, the prompt of an onboarding screen for a dog walking app generates interface designs you typically expect, ready to go as Figma frames.
From my perspective, the output is generic and what you expect from a basic UI kit (see the previous issue about taste). Despite that, the ability to generate ideas and give someone without UI design skills to have a starting place is impressive. It’s abundantly clear there is still a need for human designers to push expression and craft because AI (for now) relies on what humans have created as inspiration.
Let’s use fashion as a metaphor. A t-shirt from Target and Louis Vuitton serves the same functional purpose as a piece of apparel you wear on your torso. However, someone might buy a designer shirt because of the brand, level of quality, and expression it has. I believe software will have a similar eventuality with generative interfaces and designs crafted by humans. This is why a website created by Adam Ho is unique and special because it was created by him and has a unique expression—often imitated, never duplicated.
Technologies will evolve, but design’s impact will remain
The thought of AI can be scary for humans. Will it take away what we’re doing today? Yes, and it’s okay. As Jared Spool once said, design is the rendering of intent. I have an optimistic view of applied AI in software. Instead of being worried about artificial intelligence being more human than human, perhaps it can be assistive in a way that allows us to be more human.
The final line of Terminator 2 ends with a shot of an infinite highway with Sarah Connor narrating:
"The unknown future rolls toward us. I face it for the first time with a sense of hope, because if a machine, a Terminator, can learn the value of human life, maybe we can too."
Tools using AI
In case you're interested in AI tools, here are a few that have nice implementations of AI.
Tweet of the week
Hype links
I have a lot of questions about this…
I love Summer, BTW