AI
Artificial Intelligence
What do you think of when you hear the term?
Jarvis? Ultron? The T-800? The Entity?
First of all, if you know what any of those are you're a "nerd" and that's fine because I wrote the post so obviously, I do. However, if you have no idea what I am talking about thus far lets take a moment to discuss the reality of AI.
Humanity has been chasing the elusive "self aware" computer basically ever since we wrote the first lines of code in the 1800's...that's right the 1800's look it up. Why are we obsessed with the creation of artificial life....I don't think the character count on LinkedIn is prepared for that post so maybe we wont get into the "why" but its safe to safe we are obsessed.
So what has happened? Why are we all of the sudden celebrating our genius and being overwhelmed with AI this and AI that, seemingly in every iteration of our favorite software and tech? Because, we finally did it... kind of.
After years of talk and minor advancement, in 2022 Large Language Model AI engines made an explosive leap forward. What is an LLM? Think about Siri, Google Assistant or Alexa. When you ask any of these a question you are interacting with what is known as Natural Language Processing (they listen to your words, translate them into written form) then take those words and enter them into traditional search and retrieve answers to your questions. They then use machine learning (or the millions of collective answers they have given others to carefully tailor their response to you). Basically when you say hey Siri, its not that much different than turning to your friend in the passenger seat and saying "hey can you check what time the movie starts?" Your friend would grab their phone, open Google, type in a search, and read the response to you....thats what Siri or Alexa do. Make sense?
Now why did I tease you with Large Language Model and then explain instead Natural Language Processing and Machine Learning...because LLM is a breakthrough, and its hard to appreciate a breakthrough without some context. So Siri reads search results to you, what does LLM do that's so special? Well LLM has been fed billions of words from, articles, blogs, books, posts, tweets, reports, dissertations, computer code, music lyrics etc.. and it has learned over time how we humans write, how we create, how we articulate emotion....and the list goes on. Why is this important? Because the idea was initially to create a program that would converse with humans in a more natural way than Siri or Alexa. In order to achieve this ability the programmers let the program read millions of conversations between people in order to learn how people talk to one another, how they pause, and show empathy, how they give advice, how they tease or tell jokes etc..and after billions of words the program was pretty darn good at conversing. That was the first breakthrough. A company called OpenAI released the best chatBot in the world known as ChatGPT in 2022 (perhaps you have talked to her?). The release was met with thunderous applause. Why? Because she is smart….scary smart. Go and talk to her, I encourage you.
Was that the last breakthrough? No. That’s the thing about breakthroughs in technology, they are often organically exponential in nature. One innovation tends to spawn many. In this case, engineers and programmers realized nearly immediately that this LLM was so smart that perhaps it could handle different kinds of data…but wait (here’s the innovation) cant anything be translated or interpreted as language? For instance, you’re looking at a photo, but if you describe it to someone who isn’t with you what do you do? You use words to articulate the color or the scene, the emotion or style…..thus many different sensory data sets were translatable into words…and words can be gobbled up by LLM, and Generative AI was born.
Generative AI uses all of the same technology and power that LLM already had but is able to create visuals, or sounds, songs, and integrated scenes because underneath all of that visual and audible stimuli it is merely writing…words. Which it has been practicing for years.
So why did I spend this morning banging away at a keyboard and exceed the character limit on nearly every platform this post will hit? Because @Kyle Hanslovan (CEO at Huntress, the single greatest cyber security company in the world) posted a neat little video of the new integrated AI feature at Amazon WRITING A RANSOMWARE script for someone.
Yes. You’re right, that does seem concerning. Which is why I am posting about it. AI is awesome, and it will certainly help to propel us forward in many ways, BUT for now there are pitfalls and dangers.
Let’s highlight just a few:
Since AI can write so well, unfortunately this video Kyle posted is not unique, today a novice with no programming experience can ask AI to write malware code. What’s the risk? Access to harmful computer code can and has skyrocketed because you don’t need experience or skills to create it.
Generative AI can create pictures that aren’t real…..I am going to gently discuss this one without getting graphic. Think deepfakes, the plethora of online photos available of everyone you know…now imagine anyone with a laptop is now a photoshop expert because they can ask AI for help….see the problem?
Again with generative AI. Audio can now be created at will, you may have seen convincing videos of celebrities advertising for products or the popular Tom Cruise washing dishes in a kitchen video. With a few spoken words of your voice recorded the latest generative models can make you say anything they want…..it doesn’t take any imagination to be concerned about that.
Todays post was a long one. I hope you found it informative, I hope you familiarize yourself with this new technology and guard yourself against its pitfalls. The more collectively educated we are as a populace about the technology being developed and utilized in our world the better equipped we will be to make sound decisions about how we utilize and direct its future.
Harden the Target, stay Vigilant!
Reflections on article tagged below:
