Natuarlly Stupid
What the hell is up with AI!?
“Self Portrait”
Midjourney Prompt: Make a self portait of yourself, midjourney
“Self Portrait”
Midjourney Prompt: Make a self portait of yourself, midjourney
In this issue, we wanted to focus on Everything AI, as it has really become a big deal in everything recently. Chat GPT on GPT 3.5 came out in November of 2022, GPT 4 came out a couple mont ago, and GPT 5 will be coming out later this year. Midjourney came out just over a year ago, and version 6 is coming out early this summer. It’s not just those two platforms though. Artificial Intelligence programs such as DALL-E have been out before that, and are free to use. Stockfish and AlphaGo don’t make art or generate text, they play games. We thought that since AI is becoming such an important part of the technology, it was important to talk about specifics, how it works, and reactions to it.
This magazine took a lot of work and effort to make, so we hope you appreciate every bit of it. It was hard to organize and style it and a way that made sense and was cohesive, while also not making it unorganized. Throughout this semester-long adventure we’ve learned a lot from each other, and we thank everyone that was willing to be interviewed, as we discovered many people that work with AI do not answer their emails.
Sincerely,
David Kilday is a freshman at LASA. He is interested in Artificial Intelligence and wrote articles about the behind-the-scenes of how AI works. When prompted with the trolley problem David said that he would “stay on the track” and not get involved, causing the death of 5 orphan children rather than killing Obama and a puppy. When he is not doing homework, David often plays video games because they give him a “good challenge.” He is also planning on going to college and getting some sort of job. During his interview, he was asked what he thought would happen if all dirt was replaced with mashed potatoes, to which he responded that humans would then probably eat dirt and that the ground would squish down a little bit.
Noah Volk, preferably called Noah, is a LASA student that enjoys playing video games with friends and reading to escape. After he graduates, he wants to move to Scotland, as it’s really pretty, and get a job in computer science. In EZINE class, he wants to write about AI art and the complications with it, the people who make it, and how it works. More broadly, Noah’s opinion on AI is that “it’s really cool, it can do a lot of good things for humanity, but it’s pretty scary.” The most interesting thing about Noah is that he’s “not an interesting person.”
David KildayTheo Coleman, who is interested in rock climbing and 3D Printing will be writing about reactions to AI in our Magazine. He recently (6 years ago) cut his finger off but had a really good surgeon and recovered in time for this issue of the Ezine. Theo also has no idea what he’s going to do after he graduates because he hasn’t put any thought into it. He doesn’t have many thoughts about Basket Random, except that it’s making Traver get his work done slower and that he once was playing two games at once and was “being twice as unproductive in the same amount of time.” He also believes he has tried rice pudding before.
Traver Laudadio is a LASA student that likes to read and play video games because reading is “A nice escape from reality” and he enjoys entertainment since it is a nice relaxing thing to do after school. After graduating high school, he would like to find a job involving computer science, which hopefully won’t be overrun by AI, so he can make a living that can support a family. In this Ezine, he will write about AI’s involvement in producing text. Something interesting about Traver is that his family vacations all across the US every summer in their RV.
Quinn Kart is a freshman at LASA High School in Austin, Texas. After he graduates, he wants to find a stable job and live a nice life. In Ezine, he wants to write about how AI has changed over time, and how it is affecting people in their day-to-day life. He enjoys playing games because he “enjoys entertainment, like a normal person,” and thinks that it’s a very relaxing experience. Quinn says he is an ordinary person and there’s nothing especially interesting about him.
Quinn Kart Theo ColemanMrs. Johnson is a former Computer Science teacher that taught at Kealing Middle School for 22 years. She introduced AI into her curriculum in 2017, as she wants it to be a more commonly taught topic. She thinks that learning about AI can help a lot of people, and that it’s interesting, because she says “[AI] is a huge discontinuity in intellectual development of mankind.”
“I think that [AI is] gonna change so many things we don’t even know, we have some ideas of what things can change. But I don’t think we even have an idea of what is going to change. So I think it’s massively disruptive. And, let’s hope in a good way.” -
Anita JohnsonIt’s important to have the knowledge of how AI works. But why? Well for starters, it helps you better understand where it excels, and where it’s flawed, as well as helping you understand the capabilities and limitations of AI. Additionally, AI can help you prepare for the future, as it is becoming something that is used in more and more jobs, so learning how it works might help you work with AI in the future. Mrs. Johnson, a former computer science teacher at Kealing Middle School, said, “I think it was really important for my students to understand both the power of AI, and also the limitations.” At this point, AI is becoming such a big part of the news
“Even if you do understand the math of it, it’s sometimes hard to see what the computer has actually done” - Anita
Johnsonand daily life that it is becoming more important to learn about this than ever. So, how exactly does AI work? Well, to put it simply, it’s a lot of math due to some patterns the training model has picked up. It uses parameters to modify its output based on the patterns it has learned by using an equation. Even if you understand the math, humans still don’t really know what the equations mean. If you’re unfamiliar with AI, it can be looked at as a kind of programming,
where instead of telling the computer every single possibility and accounting for it, you just give it inputs and outputs, and it solves the problem itself. When asked about the capabilities of AI, Mrs. Johnson stated that it was changing every day, but that AI is better than humans at most things involving pattern recognition. It’s important to note that even though we really only talk about AI writing and art in depth, there are far more forms of AI and its applications.
Roberto Martin-Martin is an Assistant Professor of Computer Science at the University of Texas and is researching AI and robotics-related topics. He studied Microelectronics and Electrical Engineering and started to do research in Robotics in Germany, and then later got a Ph.D. in robotics. Martin-Martin went into the field of AI because he “was always interested to know how machines work, how they make decisions on their own and how can we make them better at that based on sensor input.”
“I’m working on reinforcement learning algorithms to learn to move the entire body of a robot that can navigate and use its arms to interact. My goal is to create a robot that can help with household chores.”
- Roberto Martin-MartinThere are three main groups of AI: machine learning, neural networks, and deep learning. Machine Learning refers to algorithms that are used to gradually improve performance. Martin-Martin says that there are many different types of machine learning, like supervised learning and unsupervised learning. Supervised learning, he states, “[assumes] we have a dataset of labeled cases.” In the words of DJ Franklin, “In order to train a model to identify dogs in an image, we’d have to feed lots of images with dogs and images without dogs into the model so it can learn to distinguish between them.” They use patterns to understand the data set they are given.
Neural Networks, a smaller group within machine learning, imitate the way human brains work. You get an input, your brain does something to that input, and then you return the output. In Neural Networks, you have the input layer, the hidden layer (which modifies the input), and then the output layer. Last but not least, deep learning is just a type of Neural Network that has more than one hidden layer. For
many AI models, people have tested and researched them enough to completely understand their capabilities and their limits. However, practically every new AI is improving on its previous version, and AI is becoming more and more powerful. Newer models though, like Chat GPT are still not yet fully understood because of their recency and their complexity. Newer language models have billions of parameters they take into account when producing an output.
AI Art is a form of AI that is trained to generate visual pieces based on a prompt. These prompts can either be extremely complicated and specific, or simple and broad.
It’s hard to judge, train, or improve an AI art model because there are a lot of subjective things that go into it, like aesthetic appeal, emotional impact, or originality. There are many benefits to AI art, but to list a few, AI art can help with starting work on an art piece if you don’t want to create one entirely by scratch, or by getting a complete work of art within a matter of minutes.
However, there are some drawbacks to AI Art, one being its knowledge of the world, like how it still struggles to do things such as make hands or people. AI art can also sometimes lack originality, as it is trained with real pieces of art, and is essentially just replicating their content or styles. As on the right, you can see an example of an art piece that was created by an AI, and, though it looks a lot like it was made by a human, it was generated by an AI.
The main way AI art is trained is by having one AI that generates art with a prompt, and another one that has to guess what the thing in the picture is. These two AIs gradually learn and progress together, making for a really strong and powerful model.
AI Writing, trained to generate a text response to a prompt, is often used to create stories or provide information about a specific topic.
It’s a lot easier to evaluate the output because you can grade it on things like coherence, accuracy, and grammar. There are many benefits to AI writing, like how it can give you an easily understood answer in just seconds, but one negative is how it can sometimes present biases.
ChatGPT, the most well-known and widely used AI text generation model, was trained with an extremely large data set on the internet, including books, articles, websites, etc. The model was then trained to recognize patterns in human speech, and, once it had a reliable database of human speech patterns, was trained to apply those patterns to input prompts and output something reliable and coherent.
People affect this Artificial Intelligence type by choosing what dataset it receives to learn from, and since the world we live in is flawed, most datasets present biases from humans, like sexism. Because the AI then learns from these biased dataset, and picks up the bias itself, because it reads text written by humans.
The more recent versions are almost always beteter due to more data, and upgraded GPT software, which stands for Generative Pre-trained Transformer, which makes ChatGPT work so well.
DJ Franklin is a Research Engineer in the Robotic Platforms Section of Southwest Research Institute. He does design work with sensors and computers to fit them into the vehicles the company creates. He got his degrees (Bachelor’s and Master’s) in mechanical engineering at Texas A&M. He got an internship at SwRI, and got a full job there after his graduation. DJ Franklin thinks AI is “fascinating to learn about, and exciting to see in action.”
“AI is fantastic at classifying images and recognizing what’s in them. Of course, we’ve all seen how amazing generative AI can be through free online tools like ChatGPT as well. Generative AI isn’t used very often for autonomous vehicles though.” - DJ
FranklinRoberto Martin Martin says that “AI is being used more and more everywhere, from making decisions, to correcting text, or predicting stock options prices.” Also a daily application of AI, self-driving cars are on the rise. Many people, like DJ Franklin, do work to further develop autonomous vehicles.
He works specifically as a research engineer at Southwest Research Institute. One of the main differences between autonomous cars and normal cars is that autonomous cars need to have sensors and cameras scattered around it, to use AI and make decisions on what to do in the moment.
Whether that be slowing down, increasing speed, stopping entirely, turning left or right, or something else, is up to the inner mechanics and AI within the car. Mrs Johnson says, “[Teslas have] been trained on data, lots and lots and lots and lots of data, not just the data that the engineers trained it with. Every time somebody gets in a Tesla, that data gets sent back to Tesla,” which is how these autonomous car companies get so much data, and are so accurate. DJ Franklin says one of the difficulties with his work is that the parts he makes “are often constrained by the available space in the vehicle. Any room [their] parts occupy is space that can’t be used by passengers and other parts in the vehicle.” He has to make the sensors as small as possible, so the passengers still have space for what they need to do.
The first thing you have to do to train an AI is to give it information to learn from. This information can be pictures, text, or any other types. The better and more information you give the AI, the better it will learn and make connections.
However, before you get data, you have to know what you want the AI to do. If you give it irrelevant information, it won’t help the AI learn and produce the desired output.
If you are using a public database, you’ll need to make sure it is good, complete, and accurate data.
Sometimes, the data will need to be labeled with what the AI should produce when it gets that image/text and has to tell you the output.
A lot of the time, the first two steps are done multiple times to create a really large and accurate data set, and more training to become more accurate to predict what the desired output should be.
Once the AI has the data it needs, it processes that data using Machine Learning and learns to find and identify patterns, relationships, and other features it can use to make predictions between the data it has. It does this over and over again to get more consistent, and the more pictures it has helps it get more and more accurate.
The output is created by using the information and patterns recognized in the processing phase and forming the output based ib the conclusions used with the Machine Learning. The output can take many different forms, such as text, images, predictions, etc. When you have finished the first two steps, you can test your AI by giving it a prompt it hasn’t seen before, and seeing what it produces. If the output is correct, perfect! You can test your AI more, or leave it how it is and use it for future projects. If your AI produced an incorrect output, then you can go back to give it data and let it process more.
So now you should now have learned the basics of how AI works: Data Input, Processing, along with Testing and Output. The diagrams hopefully have helped a little bit, and you know how each phase works. Data Input is all about giving your AI a lot of relevant, helpful, data information to build off of. Processing is about the AI recognizing patterns and relationships in the information and data it was given, so it is ready to form an output with reasoning, which brings us to our last step, Testing & Output. Testing and Output is to help you test your AI, and shows that the output it’s giving is based on the reasoning from the Processing phase, and can give you helpful information in the future.
Ever since the release of Midjourney, as well as OpenAIs ChatGPT and Dall-E, artificial intelligence has been on the rise, the center of news stories and at the heart of popular culture. What was once thought of as science fiction is now science fact, and AI art has been one of the biggest players in this field. AI art engines such as Midjourney and Dall-E have been making waves in the world of technology, and for good reason. This article will be looking at the newest features of AI art, the best image generators on the market, the story of AI in art competitions, and the complications that this revolutionary software brings.
AI art is a powerful piece of technology that only requires a prompt, with possibilities which can range from a simple three-word sentence to a paragraph of text, and a beautifully rendered image is created. Midjourney, a popular AI art generator, allows for various ways to customize your image, with various styles, versions, and aspect ratios that can make the art you’ve generated exactly what you wanted. DALL-E, the other most popular generator, allows for similar style customization, and both allow for editing or changing pre-existing images. This tech-
nology is being used in various fields. For example, Ralph Cutler, a web-designer, uses AI art to produce ideas to help his job. Ralph Cutler is a web designer for and co-founder of a company that creates websites and online tools. He uses AI art generators like Dall-E to test and create ideas to assist him in his job. He uses these programs for “brainstorming or just testing out ideas,” and usually keeps his use of AI at his job to a minimum, saying he mostly uses AI “because I do tech stuff, it’s just good to be aware of the changing technology and how things are evolving.” Some people, however, take their use of AI in their job to a more extreme level.
“It’s unrealistic and sort of immature to think you can just stop it and ban it and pretend like [AI] doesn’t exist,”
To understand the complications of AI art, it’s important to know how it works. These programs are fed images and artwork from throughout the web, scouring various websites for knowledge, sometimes even stock image sites. Because of the way that companies train and control these AIs, a few problems have arisen. First off, the images that AI generates are limited in some ways. Some of the biggest limitations that those interviewed took note of were that it can’t generate consistent characters across multiple images, for example, if you want two perspectives of the same image, there’s no guarantee that the person in your first image will be the same as the second. Additionally, most companies have restricted the creation of graphic content, it often lacks resonance and emotion, and the image it generates can be repetitive and you might find yourself unable to generate the
images that you want. AI art, despite its current flaws, has overcome a great many over the few years that the big AI art generators like Midjourney, DALL-E, and Stable Diffusion have been around. AI art was once a bit of a neat gimmick that made messy pictures that were very obviously not created by a human. At the time, it was quite difficult to craft prompts that would create clear, or even just legible images. Now, these engines can easily create beautiful images and take in a lot of information from the prompt and create a very clean picture, additionally, Midjourney, Dall-E, and Stable Diffusion have now been fed
vastly more knowledge which allows them to create better and better images. There are still flaws other than the ones initially listed. For example, they can still sometimes be overloaded with information and they don’t always generate the images that the user wants them to. In the end, AI art will always have problems to overcome, and with each solution will come a new one, but maybe, someday, AI art will be just like comissioning an artist for an artwork.
During the 2022 Colorado State Fair’s annual art competition, a piece called “Théâtre D’opéra Spatial” was entered, and won. The problem? It was an image generated by artificial intelligence, one of the first of its kind to win an art competition. This incident is one of many which brought to light the issue of: “How do we distinguish AI-generated images from human art?”. Yes, current AI-generated art can be messy or noisy, like the image below, but already AIs are starting to get better and better at creating clean and noiseless art that sometimes, even under close scrutiny, is difficult to tell from “real” human artwork. How long until that’s impossible?
ThéâtreD’opéraSpatial, courtesy of The New York Times
Next up, there are dozens of AI art websites and programs out there, which one is the best, as recommended by people who work with AI for a living, and what are the biggest flaws that AI currently faces and the ones they’ve already overcome? Find the answer to all of these questions next.
There are many different options when it comes to choosing an AI art generator for your images, but which one stands out the most as the best AI art generator on the market? DALL-E was one of the pioneers of AI art, and a generator called Leonardo provides lots of control over the images generated, but what everyone seems to agree on is that Midjourney is one user-friendly interface that allows for lots of image customization. What’s not to like?
As Samson Vowles, a YouTuber who specializes in AI art technology, puts it: “What makes Midjourney stand out is its ability to produce highly coherent and aesthetically pleasing results, with a level of refinement and precision that is difficult to achieve with other AI art platforms”
With the release of Midjourney 5, the technology and opportunity for creating AI art is at an all-time high. Some people, like Lost Space Bunny, who wished to remain anonymous, uses AI art to directly create pieces to showcase to a wider audience. Although they have been an artist for their entire life, Lost Space Bunny now also uses AI to create art. “I discovered AI imaging about three years ago when it was primarily generating nightmarish images. However, after two years, it began to create more coherent visuals, which piqued my interest.” Initially, they say that they went through a tough period, and they “had always assumed AI would target creative jobs last, but it seemed the opposite was true.”. However, they were also fascinated by the technology and used it to pursue a dream of selling art at markets.
“I believe that AI art has already begun to challenge and redefine our understanding of what constitutes art and creativity.”
“Creativity is an intrinsic aspect of being human. People will always have a desire to express themselves through drawing and other artistic endeavors,”
- Samson Vowles
Prompt: Futuristic city at nightfall
- Messy
- Noisy
- Doesn’t work well with complex prompts
- Lacking in detail
- Falls apart when you look closer
- Less knowledge
- Less realistic
- Images can be blurry or unclear
Prompt: Futuristic city at nightfall
- Clean
- Works with complex prompts
- Has lots of detail
- Holds up better when closely examined
- Vastly more knowledge
- More realistic(When asked to be so)
- Images are more defined
It’s clear from the above images that Midjourney has come a long way from its first rendition, now being one of the most revered and powerful
art generators known to the public. It’s even possible to see a massive jump in clarity and detail just looking at version 3 versus version 4.
Artificial intelligence (AI) is taking the world by storm, one of the primary forms being AI chatbots. What is a chatbot? A chatbot is a computer model that is designed to simulate a conversation with another human while being run by artificial intelligence. In November of 2022, one of the biggest advancements in artificial intelligence was released— ChatGPT. ChatGPT is able to write entire essays and text in many other forms of output(like song lyrics and play scripts) in seconds, explain practically anything, and even write code.
First off, the main software for chatbots is GPT, a Generative Pre-trained Transformer, which was developed by OpenAI. But… How does AI writing and GPT work? AI writing works by using a process called machine learning that’s based on patterns from a large dataset of writing samples. This starts by creating a dataset based on anything that humans have written. The datasets then have all punctuation
removed during processing, and the AI examines the text and begins to pick up on common patterns of writing. The AI will then be trained to be able to generate responses based on certain inputs and adapt them for different output forms, like essays or poems.
provide suggestions, and write code. These are some of the most common uses, but the other popular use is writing essays. ChatGPT is a software model that usually produces outputs in the form of an essay, as that’s the most natural way for this specific model to write. This makes it great for writing essays, which is unfortunately something that a lot of kids are currently doing in school.
ChatGPT can be used in several ways, three of which are listed on OpenAI’s main page; “Explain quantum computing in simple terms,” “Got any creative ideas for a 10-yearold’s birthday?,” and “How do I make an HTTP request in Javascript?”
Essentially, it’s saying that ChatGPT can simplify complicated concepts or terms,
While these AI writing programs are revolutionizing the modern world, they have their fair share of downsides as well. One downside of these programs is that they don’t always produce accurate information. ChatGPT, for example, was developed and trained up until September of 2021, so while it may know some things past September, it is significantly more likely to produce accurate information from before that time. OpenAI has recently developed GPT 4, the next generation of GPT that can now take the input of images, generate faster responses, and output much more accurate responses.
Image of a robotic handMany have quoted Artificial Intelligence as one of the most monumental jumps in technology of this century. A lot of people, however don’t know, what AI writing actually is.
Benjamin Andrew is a research engineer at the Southwest Research Institute, and focuses on improving “computer vision, machine/deep learning, and robotics, especially autonomous vehicles.” He decided to pursue a career with AI because he “was always interested in making things and learning how things work,” and “video games inspired [him] to learn about computers and how to program them to make useful tools or art.” He notes that if you want to have a career in AI, you should learn some programming and problem-solving, be solid with math, primarily statistics, but most importantly, actually have an interest in both of these and enjoy doing them. He also suggests that you should start early, so you don’t have to do these all at the same time.
Louis Bouchard is a Ph.D. Student, artificial intelligence Youtuber, podcaster, and head of community at Towards AI. His opinion on AI is that “it is becoming more and more powerful but is also quite simple underneath,” and currently “the more human in the loop we have, the better. Algorithms are nothing that special but engineering and social intelligence is.” As a kid he wanted to pursue a career in AI because he “loved mathematics and always wanted to become a researcher,” but now, it’s because he is “excited to work on an applied form of mathematics and productize it.” He suggests that people going into the field of AI should prioritize getting good at programming and learning to understand APIs (Application programming interfaces), as well as having a solid background in math and coding.
“I think AI is an intensely complicated and fascinating subject that intersects with so many other things humans find important. Currently, I see AI as a powerful tool that can be used for to do many great and awful things. It can also be scary or uncomfortably enlightening at times.”
- Benjamin AndrewLouis Bouchard
ChatGPT hasn’t been around long, but it’s already causing problems. ChatGPT and other chatbots are being used by kids in school to write essays for English classes, code for computer science classes, and even used in world cultures and science. Unfortunately, there are probably even feature articles for electronic magazines being written by these chatbots. Why is this a problem? Children aren’t learning anymore; they won’t know how to communicate when they get a job, and they won’t be able to write a single paper at that job. In the words of Todd Coleman, “Education must keep up with change. We must learn how the new technology works, and adapt our education system to match it because otherwise, our students will inevitably fall behind.”
Luckily, there are already programs that are able to detect whether a piece of text was written with AI. Unfortunately, they aren’t foolproof.
These programs claim to be extremely accurate, however, aren’t nearly as accurate as they claim. Additionally, most people believe that they aren’t
accurate enough to prove academic dishonesty. Benjamin Andrew believes that “this makes motivating students to want to learn to be the only real way to combat something like this,” since if the students are motivated, then they’ll actually want to do the work instead of using AI to do it for them. While education might not be the best place for utilizing AI, there is definitely a place for it to add efficiencies to the workforce. Louis Bouchard mentioned that AI will be used for repetitive writing tasks, language translation, and text editing, and Coleman said that “AI promises to exponentially speed up the process of experimentation and discovery” and we should be “ready for a massive influx of new games, books, films, and music.”
One is education, as I mentioned. Two is research (medical and scientific). A.I. promises to exponentially speed up the process of experimentation and discovery. Lastly is entertainment. Get ready for a massive influx of new games, books, films, and music.
- Todd ColemanTodd Coleman is the founder of a company that is creating conversation AI. This software will be different from software like ChatGPT because you will be able to actually talk to the AI via voice. He decided to pursue his career in AI because he had “always been fascinated by the idea of artificial intelligence,” but “never thought that [he] would actually see it in [his] lifetime.” He now realizes he was very wrong, saying that people “are already seeing the first instances of “intelligence.” What he recommends for someone that wants to have a career in AI is to have a good understanding of statistics, linear algebra, and calculus. When asked about his opinion on AI, he said that he “believe[s] it will be the single most impactful invention humanity has ever come up with. [He] believe[s] it will impact every facet of society and the human experience.”
If, in the future, your job is taken over by AI and you are no longer getting clients, what would you do?
If I’m unable to retire, I’d be forced to pivot within the financial sector. It’s been my area of expertise my whole career so I’d find a way to leverage that history and knowledge.
- Cary LaudadioCary Laudadio is a Financial Advisor and the founder and owner of Goalfusion Wealth Management. He started his career in 1999 with every meeting being held in-person, and with the help of COVID-19, is now fully online and works from home. When asked if he thinks a chatbot can do what he does, he said that “A chatbot could complete 10% of what [he does]. Every individual has different circumstances that create a unique profile and investment philosophy. A chatbot could provide certain universal advice, but when very little is absolutely right or wrong a certain amount of human subjectivity is needed to tailor advice.” Additionally, he thinks that his clients would prefer “human interaction over the chatbot. So much emotion is wrapped up in people and how they feel about money that a chatbot wouldn’t completely satisfy those emotional needs.”
Cary LaudadioPenguins are a unique and fascinating group of birds that have adapted to life in the water. These flightless birds are found primarily in the Southern Hemisphere and range in size from the tiny fairy penguin to the massive emperor penguin. Their sleek, streamlined bodies and webbed feet make them excellent swimmers, allowing them to dive deep into the ocean in search of food. Penguins are also known for their distinctive waddling gait on land, which is caused by their short legs and wide stance. They are social animals and often form large colonies for breeding and feeding, with some colonies having hundreds of thousands of penguins. Despite their cute and comical appearance, penguins are tough creatures that have evolved a number of unique adaptations to survive in some of the harshest environments on Earth. However, many penguin species are now threatened by habitat loss, overfishing, and climate change, which is why conservation efforts are so important to ensure their survival for future
January 20 is Penguin Awareness Day so here are a few interesting facts to celebrate the day. Did you know that Penguin coloring actually serves as their camouflage? Their black back blends in with the darkness of the ocean when they’re floating in the water and the penguins white bellies look like the bright sky when viewed from underneath. Their coloring disguises them from predators like leopard seals and helps them catch prey like fish and squid. Another way penguins survive in their environment is by having the ability to drink saltwater but process out the salt. Penguins have a supraorbital gland located above their eye that removes salt from the bloodstream. The excess sodium then comes out through the bill or by sneezing! Penguins aren’t just cute and friendly, they are also smart, adaptive birds.
23.4% Incorrect
76.6% Correct
In the form, I polled people about why they picked what they picked, and a couple of my favorite respones are below.
“No AI in the world uses exclamation marks for normal paragraphs.” - Yale Xie
“[The first one] is more matter-of-fact, and the 2nd one has more character.” - Miller Williams
• Usually, Chatbots don’t use exclamation marks.
• Chatbots are coded to write essay responses, so they’re responses are ususally going to be in that format.
• There are lots of websites online that have technology that’s able to differentiate between AI tect and human text, so you should use those if you ever want to be sure.
Most people are unaware of the numerous ways that AI can be used in our everyday lives. From our personal phones to stock markets making millions of transactions every day, AI is the driving force behind many technologies and activities we take for granted. “Once people get used to the latest capabilities of AI, they forget that it’s AI,” says Gartner Analyst Svetlana Sicular.
Gartner is a research firm that advises companies on technology trends and adoption, and Poshmark is an online marketplace for new and second-hand clothing and accessories. Three leading AI experts—two from Gartner and one from Poshmark— weigh in on the many surprising ways AI surrounds us in our current day-today lives.
Svetlana Sicular researches AI and gives advice to companies on how to effectively utilize it in a corporate setting as a VP (Vice President) Analyst at Gartner. With all of the recent hype around AI, companies are often left wondering what’s possible and what’s not. Sicular states, “I speak with
enterprises about how they implement AI in different circumstances…I help them find the right strategy for a specific situation.”
Sicular believes that the place where we encounter AI most often is on our phones. For example, things like autocorrect, speech and image recognition, photography, and social media feeds are all driven by AI. Furthermore, she makes a point that early COVID-19 vaccines used AI to get to market as quickly as they did, saying that “[AI] shortens the development cycle because it eliminates some dead ends. You can get through the stages much faster.”
Barkha Saxena is Chief Data Officer at Poshmark, an online retail company. Her company uses AI to determine the prices and discounts of products, as well as marketing promotions. In her words, “In my role, I manage the data from end to end. A big part of it is machine learning, AI, and data science.”
Daily examples of AI she cites include more familiar things like Netflix and YouTube recommendations, as well as less common applications like more intelligent medical diagnoses and autonomous vehicles.
Rita Sallam is a Distinguished VP Analyst at Gartner. She’s interested in how data and analytics are being impacted by AI. “ChatGPT has been a huge market catalyst for AI.” This is because ChatGPT substantially increased the availability of AI to the public. She considers AI to be found everywhere, including, but not limited to, GPS, healthcare, and stock trading. Take it from the experts—AI is all around us. From everything we watch, to how we interact, to how our data is processed around the globe. As more and more AI gets embedded into daily activities, the less people think AI is something novel and different. It just becomes part of how things are done. AI is here to stay!
“Once people get used to the latest capabilities of AI, they forget that it’s AI.” - Sicular
AI is evolving at such a rapid pace that it’s hard to imagine what the cutting edge of AI will be in 5 or 10 years and what issues could arise. Research firm Gartner predicts: Through 2026, more than $10 billion will have been invested in AI startups that rely on foundation models (large AI models trained on huge amounts of data).
What this means is we don’t yet know what the outcome of this investment will be. In Sicular’s words, “The overall hype around generative AI is through the roof. I have never seen AI go viral this quickly.”
Gartner believes technologies follow a common pattern called a “Hype Cycle” where after reaching the “peak of inflated expectations”, technologies go through a sort of reality check in the “trough of disillusionment” before
emerging in a more realistic “plateau of productivity.”
Therefore, if AI is at the peak of its hype today, what lies ahead is a fair amount of disappointment until people realize what it can and can’t do well.
The NYT says AI will move from just primarily text to getting very good at images, voice, and video. There are already copyright lawsuits from artists and musicians who claim AI is stealing their intellectual property.
Sallam states that the concept of “deepfakes”, taking videos of people and using AI to superimpose someone else on top, including their exact voice, “Has a lot of potential for criminal and illegal activity”
NYT says AI companies: “hope to eventually build what researchers call artificial general intelligence, or A.G.I. — a machine that can do anything the human brain can do.”
It is not yet clear if this means AI will make people more productive or replace workers, but experts believe that AI is most likely to disrupt jobs with repetitive and less creative tasks,
including things like tech support. How the future of AI evolves will also depend on cultural issues and how they are perceived. Sicular observes, “In Japan, people love robots, while in the US people associate robots with the Terminator.”
In the end, the future, like always, could happen in any way possible. AI might take over humanity, help it, befriend it, or some other possibility. We’ll just have to wait and see.
“[AI will improve] at least two to four, probably three to five times more than what it is today.” - Saxena
The roots of AI began in the 1950s, but it took many key developments scattered throughout the last half-century to make it a cuttingedge, all-pervasive technology.
AI has been around for a long while, but it took many things coming together to make it happen: An explosion in the availability and sources of data, massively increased computing power, and a usable interface to make it accessible to the general public
There’s much debate about when AI started; however, many people would argue that Alan Turing’s “Turing Test” began a crusade to see if humans could build a machine that would
respond like a real person. Many early “machines” were built to play games like chess or checkers and see how they compared to human players. One of the earliest examples was SNARC, an AI comprised of 40 “neurons” that could solve a maze automatically.
The AI experts unanimously agreed that the development of AI happened in steps. Sallam mentioned a major milestone that happened in the 1960s. “Silicon chips were the first stepping stone to enable the ability to compute in an automated way,” she said. Not long after, Wabot-1 was developed by Japanese researchers in 1967. It could communicate in Japanese and measure distances and was considered the first anthropomorphic robot.
Another significant milestone came in 1982 with the Hopfield Neta recurrent neural network that made machine learning much faster. Both Sicular and Saxena pointed out the need for data to train these models,
and the advent of the internet led to such an explosion of data.
By the mid-2000s, many industries are exploring building models with data. For example, in Financial Services, high-frequency trading (a.k.a. algorithmic trading) takes off and neural networks are first used in fraud detection.
Another pivotal moment is the development of ImageNet, a database of images, and the ImageNet challenge, a university competition where people try to classify images with the lowest error rate. In 2012, Sicular noted, “Alex Krizhevsky won the competition with
“...It makes this dramatic move that people don’t want, and that’s the beauty of AI.”
- SicularBarkha Saxena
his algorithms which are now known as AlexNet. He used deep neural networks, which…require a lot of computing power and they require a lot of data. He used graphical processing units (called GPUs) for deep learning… and that started the hype in deep learning.” AlexNet blew all other approaches away with a 16% error rate, while the next closest contenders were in the 2535% range.
In 2017, the AI AlphaGo was made by the company DeepMind and was designed to outplay humans in the game Go. Its first iteration, AlphaGo
Lee, was named for its iconic match against world-class Go player Lee Sedol. Soon after, DeepMind made an interesting choice. They reset AlphaGo’s training data back to zero and taught it again. Only this time, it learned solely by playing against itself. This version, called AlphaGo Zero, was able to outperform AlphaGo Lee in under three days of training and went on to become the best Go AI ever created, by far. Sicular observes: “The AlphaGo algorithm made a move in one of the games, it’s what’s known as move 37. But the algorithm doesn’t know [that the move is uncommon], so it makes this dramatic move that people don’t want, and that’s the beauty of AI.”
Recently, the famous AI known as ChatGPT emerged. Made by OpenAI, it truly made the prowess of artificial intelligence available to everyone. Also, something to keep in mind is the amount of energy needed for much of this computing infrastructure. “Training GPT-3,
which is a single general-purpose AI program that can generate language and has many different uses, took 1.287 gigawatt-hours, according to a research paper published in 2021, or about as much electricity as 120 US homes would consume in a year,” states a Bloomberg Article by Josh Saul and Dina Bass.
People have always had a hard time accepting decisions made by data and machines until there is enough understanding and value seen from it. As Saxena says: “I think that ChatGPT also has to go through the same evolution, that people have to start fully understanding and accepting it.”
“ChatGPT also has to go through the same evolution, that people have to start fully understanding and accepting it.” - Saxena
SNARC was made in 1951 by Marvin Lee Minsky. It was made with 40 artifical neurons and could find the exit of a maze, similar to how a rat can.
1951
1967
WABOT-1 was made in 1967 by Ichiro Kato and was a huge step in AI development. It could communicate with a person and measure distances to objects.
The Hopfield Net was made in 1982 by Dr. John J. Hopfield. It was a new type of recurrent neural network that allowed for much faster machine learning.
1982
Other notable AIs include:
Logic Theorist, 1956: Used a search tree to do logic and math
Shakey the Robot, 1969: The first robot that could sense things around it and react
BKG, 1979: A backgammon robot that succeeded in defeating the world champion
AARON, 1985: A series of programs capable of creating artistic images
Cog, 1993: MIT’s attempt at making a robotic humanoid child in five years
VaMP and VITA-2, 1994: The earliest self-driving cars, able to drive at up to 80 MPH
Furby, 1998: The first time AI reached a household level of integration
Roomba, 2002: A vacuuming robot that used AI to find the shape of the room
AlphaGo Zero, 2017: Played the board game Go, and could beat the best human players
Midjourney, 2022: The most advanced AI image creator, capable of making amazing art
Kismet was made in 2000 by Dr. Cynthia Breazeal of MIT. It had sensors that enabled it to recognize human emotions and replicate them using its face.
Atlas was made in 2013 by Boston Dynamics. It is one of, if not the most advenced AI in terms of physical movement, able to do many things, including parkour.
ChatGPT was made in 2022 by OpenAI. It is commonly known due to its ability to generate text of any kind, from answers to game questions to full essays.
On the last page there are the results of a survey I took regarding Artificial Intelligence and if its a good thing for society. Meh was the majority, then yes, the no. But why is this?
To learn about the reasons people have different opinions on AI I interviewed three people, one for AI, one against AI, and one neutral about AI. All three of these interviewees work in the same industry, the game industry, so I could learn about their differing opinions on the same matters. I Interviewed J. Todd Coleman, Jon O’Neal, and Jesse Scoble, who all work with and around AI.
Edited by Theo Coleman‘Meh’ was the most popular response to the poll by a lot, but why is this? Maybe everyone who filled it out was just lazy and didn’t want to put effort into their response, maybe some people like some parts of AI, like AI writing, while also hating others, like AI art. There are a million different reasons people might have mixed feelings on artificial intelligence, but there is one argument that sums it up pretty well. According to J. Todd Coleman, an ex-game designer from Austin, Texas, “It really depends on how people choose to use it.” Just like any tool it “can be used for good or for bad.” Take Photoshop for example. It’s an amazing art tool used by many talented artists to make beautiful art. Additionally it’s very helpful for quick photo editing like removing backgrounds from photos. However Photoshop is also used for a lot of bad stuff, like catfishing and forging images. Artificial intelligence is like Photoshop in this sense, but “the potential for good and the potential for bad are orders of magnitude more impactful.” People don’t get mad at Photoshop though, mostly because you have to pay for it, causing it not to be readily available for anyone with malicious intent. Programs like DALL-E and MidJourney, however, have an extended free trial period before having to pay. Unless these companies start putting more effort into stopping people from doing bad things with it, AI is going to continue to
“It really depends on how people
choose to use it.” Just like any tool it
“can be used for good or for bad.”
-J Todd Coleman
J Todd Coleman
Although ‘meh’ was the most common response in the poll, ‘yes’ wasn’t far behind. The idea that AI is good is entirely based around the idea that it is going to progress the world incredibly quickly by optimising tasks. This opinion on the matter is more prevalent in youth because they don’t have to deal with jobs, and the idea of AI taking jobs doesn’t scare them. AI taking jobs also doesn’t scare Jon O’Neal, a video game artist from the Austin area, but for a completely different reason. “AI is coming, there’s no way we can stop it” he said, “and so those who are ready for [AI], I think stand a better chance to kind of weather that storm.” Jon has known about artificial intelligence for over a decade, but up until recently he didn’t pay much attention to it. That was until art programs like DALL-E and MidJourney started coming out. It was “about two or three years ago, when we first started hearing about MidJourney…like, whoa, okay, now they can start doing some interesting stuff.” As an artist, he knew that he could either learn how to use it or get left behind. Now he works for a company called Monumental, where he works to figure out how AI can be used to enhance the speed of workflow. When making game art, there are a thousand different hurdles you have to go over, including concepts, rigging characters, greyboxing, user interface, and more. Now “that’s what I’m trying to figure out is where does all this AI fit in?” to make this process easier and quicker. A good example of this is in concept art. “You can take a team of like two or three concept artists, they make four or five different types of concepts. Then they run it through something like MidJourney, but in a closed system. And then they generate 50 or 60 different concepts just based on their own art.” This is different from normal AI art generation because it uses art the artist made as inspiration for the generated art, instead of random artists on the internet. This way artists aren’t losing jobs, but instead being more efficient by using AI as a tool. “If you can manipulate [AI] in such a way and augment your abilities, I think there’s some value there, I really do. But the key word, there’s a few key words in here from an artist standpoint, augmentation, not replacing”.
“If you can manipulate [AI] in such a way and augment your abilities, I think there’s some value there, I really do.”
-Jon O’Neal
Jon O’Neal
Although ‘no’ ranked much lower than the other two options, it is the far more popular opinion for the older demographic. The main argument that AI is a bad thing for society is that it’s taking people’s jobs. AI art programs, such as Midjouney, are making professional-grade art that would take artists days or even weeks, in seconds. Now, take game companies for example. Why would they hire a team of artists to make concept art for their games, if they could just hire one guy who’s good at using AI, especially considering that these programs can imitate the art style of any person, game, show, etc. This, however, is not totally correct according to Jesse Scoble, a Canadian game designer, who believes that the rise of Artificial Intelligence will only affect new people coming into the working force. Take writing for example, companies can hire five talented writers coming out of college looking for entry-level positions, or “they can start giving those tasks to non-writers, basically, people who are not trained in English, and instead AI. And it won’t necessarily be great text, but it will be okay, techniques that will be, you know, good enough.” Also a company would only need one of those because its writing is good enough, and a lot faster. Scoble thinks that “people who have kind of the experience in a field for years, you know, hopefully won’t have their jobs taken, but I think for juniors coming up, sometimes it’s really good to have these, you know, simpler tasks, these easier jobs to do for them to learn and to kind of get their feet wet. And it [AI] may take away a lot of those things.” This also brings up the age-old paradox: ‘How do I get experience without a job, and how do I get a job without experience’. Now with big companies deciding to hire AI specialists to do these simple tasks, “it’s
“people who have kind of the ex -
perience in a field for years, you
know, hopefully won’t have their
jobs taken” -Jesse Scoble
When prompted with the question of ‘Is AI good’ most people, surprisingly, had no preference and answered ‘meh’. ‘Meh’ beat ‘yes’ by a staggering 7%, and ‘yes’ destroyed ‘no’ by 47%. There were two main mentalities for the majority of the people who took the survey and answered ‘meh’. The first mentality is that they don’t know enough about it to decide if Ai is good or bad. All the responses like this made me realize that people think Ai is good or bad based on how much they know about it. At first, people have no opinion on it because they don’t know much about it, then as they learn about it they either think it’s really good or really bad depending on what they hear about it. If they hear that it’s helping students study effectively they would think it’s good, but if they heard it’s making it easier for students to cheat they would think it’s bad. Then as they learn even more about it they hear both sides of the argument and fall back to ‘meh’.
This leads to the second mentality which is that there is both good and bad coming out of Artificial intelligence. Ai is helping advance technology really quickly and make lives easier through programs like ChatGPT, but these same programs can be used for malicious intent, like cheating on homework. All the positive responses that we received from this survey said about the same thing, that Ai is helping us be more efficient, progress faster, etc, etc, except they left out the second part about the negatives of AI. The negative responses did the same thing except flipped. This proves the fact that AI is a neutral thing and that it is a good thing or not depending on how we use it.
“I think it will come eventually no matter what, so why shouldn’t we get used to it now? ChatGPT helps countless students understand topics that they would never even have considered before. And Call-E has more use cases than you could imagine”
“AI has already been used for deception and I think people will continue to use it for that”
“AI has already been used for deception and I think people will continue to use it for that”
“Jobs are getting taken by AI”
“It has no way of knowing if it is doing something that will help or hurt humanity. It could do something”
“There are many advantages to AI, which gives us things like self-driving cars, but there are also some negativs. Its scary how good AI has become”
“Idek, this survey kinda sucks ngl”
“I dont know much about it”
“It is able to increase the efficiency of people and increase the growth of technology.”
“It has endless applications, especially in the medical feilds”
“I think AI could be useful in some cases, like making difficult jobs easier for us, but people being too depending on it could be harmful”