Am I in a long line of people that have written about Artificial Intelligence (AI)? Yep. Is there a longer line of AI generated content? Possibly. I waited a while to write about it because I wanted to see how it would evolve as well as how people use it. It’s still rapidly changing, this may be out dated in a month or less, if so, I’ll be back to update.
What is AI?
AI is a general term that has many applications. The hot new model is Generative AI – such as ChatGPT, MidJourney, and so on. It takes content and regenerates it into something “new”. It’s not actually new of course, just a new combination of elements and styles. Machine Learning is another branch of AI that primarily analyzes large volumes of data to find patterns and trends. This article is about Generative AI.
As with any tool, it has it’s pros and cons. As with many topics today, it has the potential for extreme good and extreme bad. Let’s take a look at some of each:
Benefits of Generative AI
Generative AI can do a lot of helpful things to boost human output (some might argue that’s not necessary, but that’s another blog post!), boost our creativity, and simply improve what we produce. Some examples are:
- It’s fantastic for brainstorming ideas
- It’s helpful at getting past writer’s block, or finding a better way to say something
- Bing AI in particular, is great at showing references for further info. I LOVE this.
Cons of Generative AI
There are some pretty serious cons to go with those benefits.
- It is built on people’s creative and copyrighted works – usually without attribution, much less payment for licensing it. This is a huge negative for me, and one that has the potential for radically changing the cost or existence of generative AI. The New York Times and many authors have filed suits for copyright infringement. In response, OpenAI, the creators of ChatGPT said “it would be impossible to create tools like its groundbreaking chatbot ChatGPT without access to copyrighted material“. Well. Maybe they should have thought about that before creating their tools.
- The potential for Deep Fakes is here. Sometimes this is used for fun, but the potential for harm in legal issues, elections, and harassment is very real. Ethan Mollick gives a great example by creating a fake video of himself. While there, read all of his stuff, he has many more examples of the good uses of AI.
- This is more philosophical on my part, but I believe communication is a valuable skill. It’s part of what defines us and makes us human. Learning to express our thoughts, emotions, and wishes clearly and kindly is critical to our quality of life. There is a fine line on using generative text to help us do that and letting it do all of our talking for us. I’ve already tried to have conversations with people about something they wrote and failed because not only did they not write it but they didn’t even read it before posting it. On top of that, it wasn’t even a subject that they knew much about, yet they passed that information off as their expert opinion and work. At best this is like someone using a 10 year old photo on a dating site, at worst it destroys your credibility.
- AI often makes things up. Whether it be mutant humans or just misinformation do make sure the information provided is correct – and not just that it seems correct. Most of the models learned from what was on the internet, and I think we can all agree that not everything on the internet is true! In some cases it could be mistaking a work of fiction with truth, in other cases it can be learning from bad actors deliberately posting misinformation.
- Bias is amplified. Again, since it is learning from what is on the internet, and much of that is written with bias – so is the output. If you are using one of the text generators, make sure that it is written for everyone.
- Privacy, or lack of is a concern. The text prompts you upload might be used to further it’s learning. Think carefully before uploading private information, personal details, and proprietary information about your business or your clients. Read the terms of whatever program you use carefully and check out all the settings.
The last three points are improving. Image generation in particular is much better, but the occasional 3-eyed person will show up if you don’t look carefully. Always proofread and alter the output as needed.
The first two are not easily solved.
0 Comments