How To Regulate ChatGPT In 10 Steps
It’s time to talk about ChatGPT and Machine Learning and how they can make or break the internet. ChatGPT and tools like Dall-E expose several problems and issues with the internet today and dangers for the future. It risks discouraging creative, original thinkers, and aspiring artists. But simply banning it won’t work. We need to get creative in thinking about how to inspire that these tools are used the right way, to inspire critical thinking, and improve the quality of content online. This is how we can start.
How To Regulate ChatGPT
- AI-generated content must list and recommend possible sources and inspirations online so that people can learn about possible inspirations and where different ideas come from and can make speculations about their value and validity.
At the moment, as we cannot see possible sources and inspirations to different prompts, images, and AI-generated content, we can’t trace or track the validity of the content. This makes critical thinking impossible – statements have to be understood by their face value not the ideas origin or the thought process behind it.
- AI-generated content needs to become more personalized. There should be a specific style or watermark/brand identity in the tool to ensure that people can track it’s origins and that it’s easy to identify if content is AI-generated or not.
By giving all AI generated content its own unique brand identity, we ensure that works generated using these tools fill a specific niches and purpose, and avoid a movement towards standardization. There is a significant risk that AI and machine learning tools drive a movement online towards uniformity, where all images and articles online become increasingly similar.
3. Use machine learning to tackle machine learning. Machine learning algorithms could compare an online article to other existing articles online and give a rating from 0% to 100% how similar this article is to other articles online, and could be used to rate content and recommend contrasting, or unique viewpoints.
If we, when browsing a website online, could get insights into how original an article is, what it’s potential sources are, and if the article appears to have an inherent bias, and if machine learning could link us to contrasting or alternative viewpoints, we would improve critical thinking online.
Protect artists and creators from ChatGPT, Midjourney, Dall-E
4. A significant portion of the profits generated using machine learning should be used to promote and support aspiring artists, for example by providing people access to music, learning tools, or education for students.
Machine learning should inspire and support original creators and aspiring creators. The only way machine learning can continue to grow and improve is if we continue to grow and improve our own content and articles. Machine learning is dependant on the work of original thinkers, and creatives, and if it is to continue to exist, it needs to support initiatives which lead to an increased quality of content online.
5. Creating your own personalized AI or machine learning tools
People should train and create their own AI and tools to aid them in their own creative process, tools that remember your unique tone of voice, and how you think, and can help support you in your creative process, can help you search through previous articles you’ve written. Instead of relying on standardized archives, these tools should focus on your personal tone of voice and help you grow in your own writing and the kind of art or content you love to make.
6. Decentralise AI
It’s integral that we avoid the creation of centralized AI tools like a Google Bard or ChatGPT taking over the world or market. We can’t risk one company alone owning and controlling and using all the information published online. We need to ensure that these tools remain open source and easy to use by anyone anywhere in the world.
7. Make it illegal for online companies to own your data without your consent
In the future, people should have their own encrypted digital wallet or passport, containing all their online information. People should be able to share on a need-to-know basis, with websites they engage with online. People need to be able to withdraw consent and delete data. Similarly, you should own any content you create online, including videos, articles, and images, and more.
8. Instantly see which companies and websites that have your data, and what data they have on you.
From your wallet, instantly be able to track who you are sharing your data with, and how.
9. Promote organic ranking above technical SEO
Machine learning can too easily game technical SEO and can use technical SEO to outrank human generated content online. Technical SEO forces writers to write increasingly standardized, similarly formatted articles online, for ranking purposes, driving the internet to become more and more uniform. Instead of ranking and recommending content based on technical SEO, we should primarily use technical SEO to list or describe what an article is about, and rely on organic ranking factors to determine the relevance of an article.
10. Use machine learning for learning and personal growth
Machine learning should be used not to replace human workers but rather to inspire and empower human workers in improving the quality of their content. For example, by describing an artists painting and showing them potential inspirations, or helping them generate or test concepts or new ideas, or showing potential improvements and changes they could make. Ideally, machine learning and AI has the potential to learn from your current level and work from who you are and what you do and to empower you to grow and improve in your work.
- When asking machine learning to create a new plugin, instead of simply writing the code, machine learning could work from the users current level and help them improve their own code and writing.
- Instead of writing copy for a writer, the machine learning could analyze the users writing, and help them improve the quality of their copy, or help them notice bad writing habits, or help them research alternative viewpoints or different arguments.
- When trying to learn a new skill, AI can work as a teachers assistant, helping users learn about a new topic or practice a topic by creating flashcards, analyzing their writing, or improving their content.
11. Preventing cheating in schools
Many schools are sounding the alarm about ChatGPT being used to write essays. This can’t be stopped. However, grading criteria can be changed. Students need to be encouraged to develop their own unique, distinct tone of voice, and to actively not just describe a subject, but their own thoughts and ideas on a subject, meaning bland or uniform essays that provide stereotypical, simplistic views get graded lower. Students also need to be requested to provide sources and to critically analyze their own arguments. Teachers need to come up with more creative essay ideas and review criteria to ensure that students can’t cheat too easily and that students are encouraged to do their own research and write in their own tone of voice.
Learn more from my conversation with Leon Tsao on AI.
These are just my initial ideas after a quick brainstorm. What are your suggestions or thoughts?