Should you use AI?

So you're thinking about using AI! This is a major topic, and there are a lot of conflicting viewpoints, as well as many concurrent ethical arguments around the use of these technologies. I'll ask you a series of questions to help you decide if you should use AI. Along the way, if you hit a dead end, I'll explain some of my reasoning! First, Are you planning to use AI? - [[I never plan to use AI!|You're not going to use AI]] - [[I have to use AI at work|You're going to OR have to use AI]] - [[I'm planning to use AI|You're going to OR have to use AI]]

(b4r:"dashed")+(b4r-size:2)+(b4r-colour:green)[(text-colour:green)+(bg:(hsl:90,0.8039,0.5,0.45))[''Oh you're not planning to use it! That's just fine!'']] You're not alone. There are growing communities of folks who are abstaining from the use of AI for recreational, business, and creative means. There's no harm in not using it, despite the hype or people trying to convince you otherwise! I don't think our responsibilities end when we decide we're not going to use AI, though. Out of curiosity, do you know much about how Generative AI works? - [[I don't know much about how $usecase works]] - [[I feel very informed about how Generative AI works|Which one?]]

(You said you either are going to or are curious about using AI). Right now in our culture, when we refer to "AI," we're actually referring to a bunch of different computer technologies in a trenchcoat. Do you know which kind of algorithm / computer program / machine learning tool you're going to use? [[Yes! |Which one?]] [[I don't know; which are there?|What do you want to use AI for?]] [[No, but I don't mind|Don't use it then!]]

That's okay! (text-colour:white)+(bg:red)[(b4r:"dashed")+(b4r-size:2)+(b4r-colour:red)[''Since you don't know much about how AI works, I would recommend you please don't use it.'']] AI is as revolutionizing as a car. It offers all kinds of freedoms to adults who are safe and licensed to use it! But if you don't know what to expect from your car, even on a basic level how it works, and haven't practiced driving with supervision, it can be a really dangerous machine. You may begin to think you're invincible and hurt yourself or others! It's also really important that people know how these work, even if they don't plan on using them. As a citizen of your country, you may have lawmakers and leaders in your community that are going to make decisions about AI's presence in your world.

(You said you know which AI you're using) Which kind of "AI" are you planning to use, either at work or by yourself? - [[Machine Learning Applications]] are much more specfic applications of machine learning and deep neural network programs. They will be trained on very specific datasets and maybe have unique architectures that enable them to handle one or two very particular tasks, like unfolding proteins, identifying the presence of cancer in an X-ray, or analyze trends across massive datasets that are invisible to the average worker. - [[An AI Coding Tool]] might be a specific, proprietary chat to code interface that allows someone to plug in or connect a code base or Github account and generate working code in a variety of languages. These applications can troubleshoot code, generate new code, or reform file structures, etc., making them powerful, but often risky when given too much power to see and edit databases. - [[The big Chat models|One of the big Chat models]] are massive multimodal large language models that don't just talk back and forth, but also have big interactive capabilities. These models can take a mix of actions, like read documents, code, generate images, perform internet searches, and interact with a browser window or websites. - [[A CharacterAI/Companion]] is a chatbot specifically tuned to play a character with personality traits, backstory, memories, tone of voice, or appearance, often an existing fan character. These bots might be trained on fan fiction that people wrote with these characters, with book or movie scripts, or along the lines of certain tropes or with social media accounts, especially if the character is a real person.

(b4r:"dashed")+(b4r-size:2)+(b4r-colour:red)[(text-colour:red)+(bg:(hsl:90,0.8039,0.5,0.45))[''Don't use it then!'']]

(set: $usecase to "Machine Learning") (set: $useselect to true) You picked "Machine Learning Applications." Do you know how your model works? In other words, do you know: - What the input/training data is and where it came from; - How the model is trained and set up; //and// - How the model is meant to be deployed in real life? [[Yes!]] [[No.|Don't use it then!]]

(set: $usecase to "Code") (set: $useselect to true) You picked "An AI Coding Tool." Do you know how your model works? In other words, do you know: - How the model is trained and set up to be used; //and// - How to get the best result from the model from the get-go by responsible and smart prompt engineering? - How to troubleshoot in the appropriate software to solve the problem? [[Yes!]] [[No.|Don't use it then!]]

(set: $usecase to "Chat") (set: $useselect to true) You picked "One of the big chat models". Do you know how your model works? In other words, do you know: - What the input/training data is and where it came from; - How exactly the model is trained and set up for users; //and// - How to get the best result from the model from the get-go by responsible and smart prompt engineering? [[Yes!]] [[No.|Don't use it then!]]

(set: $usecase to "CharacterAI") (set: $useselect to true) You picked "A Character AI/Companion." Do you know how your model works? In other words, do you know: - How the model was trained and set up to be used; //and// - Who is behind this model/program/app/software? - How binding the terms of service for this is? [[Yes!]] [[No.|Don't use it then!]]

(set: $proceduralKnowledge to true) Great! I only have a few more questions, since it seems like you already hava good grasp of what the tool is you're using and how to use it, including what you can expect and not expect from the model in terms of data and output. [[I actually don't relate with that statement. Really, I just want to use it and I don't mind not knowing how it works|Don't use it then!]] First question: Are you going to give your program access to any sensitive information, like - Personal or private information about yourself or another person? - Sensitive or proprietary information about your company? - Complete access to your computer, hard drive, or file structure? [[Yes|Is it secure?]] [[No|Own Server?]]

You said that you plan to use some private or identifying information as input for your model. Do you know for sure that you have the right to work with this data, and that it will be held securely and safe to use in the AI architecture/model that you're working with in this project? -[[Yes|Undoable?]] -[[No|Don't use it then!]]

That's good! Next, are you running your model on a personal or work computer, or is the computation happening in the cloud or at a remote access site? Really, what I'm asking about is security: Do you know for sure that you have the right to work with this data, and that it will be held securely and safe to use in the AI architecture/model that you're working with in this project? - [[Yes|Undoable?]] - [[No|Don't use it then!]]

Right now in our culture, when we refer to "AI," we're actually referring to a bunch of different computer technologies in a trenchcoat. - [[Machine Learning Applications]] are much more specfic applications of machine learning and deep neural network programs. They will be trained on very specific datasets and maybe have unique architectures that enable them to handle one or two very particular tasks, like unfolding proteins, identifying the presence of cancer in an X-ray, or analyze trends across massive datasets that are invisible to the average worker. - [[An AI Coding Tool]] might be a specific, proprietary chat to code interface that allows someone to plug in or connect a code base or Github account and generate working code in a variety of languages. These applications can troubleshoot code, generate new code, or reform file structures, etc., making them powerful, but often risky when given too much power to see and edit databases. - [[The big Chat models|One of the big Chat models]] are massive multimodal large language models that don't just talk back and forth, but also have big interactive capabilities. These models can take a mix of actions, like read documents, code, generate images, perform internet searches, and interact with a browser window or websites. - [[A CharacterAI/Companion]] is a chatbot specifically tuned to play a character with personality traits, backstory, memories, tone of voice, or appearance, often an existing fan character. These bots might be trained on fan fiction that people wrote with these characters, with book or movie scripts, or along the lines of certain tropes or with social media accounts, especially if the character is a real person.

(set: $secure to true) Ok sure, so now that we've made sure that the data you're working with is safe, next… A: If you or the program you're working with does something wrong, can you reverse it? B: Do you and your team have the expertise and familiarity to tell when something isn't working the way that it should? - [[Yes to A and B|Accountability]] - [[Yes to A, but not B|I don't know much about how $usecase works]] - [[Yes to B, but not to A|I don't know much about how $usecase works]] - [[No to either A or B|Don't use it then!]]

(set: $undoable to true) Ok, so at least you can tell what's wrong with it and/or undo any problems that occur. One more BIG question: Will you take accountability for what happens if you go through with this? - Do you take criticism well? - Do you have a contingency plan? - Do you have an authority to report to? - Will you not keep your work, not less your mistakes, a secret? [[Yes]] [[No|Don't use it then!]]

(set: $accountable to true) (b4r:"dashed")+(b4r-size:2)+(b4r-colour:green)[(text-colour:green)+(bg:(hsl:90,0.8039,0.5,0.45))[''I think you're good to go!!'']] AI is as revolutionizing as a car. It offers all kinds of freedoms to adults who are safe and licensed to use it! But if you don't know what to expect from your car, even on a basic level how it works, and haven't practiced driving with supervision, it can be a really dangerous machine. You may begin to think you're invincible and hurt yourself or others! It's also really important that people know how these work, even if they don't plan on using them. As a citizen of your country, you may have lawmakers and leaders in your community that are going to make decisions about AI's presence in your world.