I've been doing my blogging on HIVE as I am able to earn crypto for my effort.
I want to experiment with AI programs like Google.Bard and ChatGPT. Since HIVE was actually paying me, I do not want to republish AI generated content on HIVE.
Since blogger doesn't pay me. I thought it might be a better location to discuss this experiment.
My first experient was to ask ChatGPT to List the Top Ten Attractions in Utah Counties? This question was easy as I could judged the results objectively by seeing if the location was in the county or not.
Both AI models failed. The most humorous failure was that Google.Bard thought that the Utah Shakespeare Festival took place in Washington County. It takes place in Iron County. A program named "bard" should know about The Bard!
For this post I asked Google.Bard and ChatGPT to list the pros and cons of using AI queries in a blog post. I will discuss the response after this paragraph and reproduce the answers at the end of this post.
Since Blogger is owned by Google, I decided to start with Google Bard's response
The "pros" section of the response comes off like an advertisement. It says that AI can increase traffic and engagement.
I think this response is wrong. While people enjoy actively engaging in chats on the AI web site. I don't think that people enjoy reading republished chats.
I created the top half of this post. You can engage with me about this section. But the bottom section is just dead. It was generated by a computer. You can go to Google.Bard, but you can't engage with the republished conversation
The Google.Bard has two nonsense answers about brand awareness and lead generation. I seriously doubt that anyone is more likely to engage in a monetized activity because there is AI content on this page.
The "cons" listed by Google.Bard are somewhat bizarre. The first con makes some sense. It says that a person should read the chat before republishing it.
But I guess there are some people who will ask an AI to do their homework and not bother reading it.
The second and third response are weird. It says that it is costly to transcribe and and edit the chats.
Let's see. To republish a chat one must highlight and copy the chat and then paste the chat. That is three steps. NOTE: There is ofte a copy button on the page reducing the effort to one action. The second con is that it is expense to hire someone to copy and past text. Here is the Bard chat. I will look at the chatgpt response here
What are the pros and cons of republishing an chat with ai on a blog or web site?
I asked the same question of ChatGPT: What are the pros and cons of republishing a chat with ChatGPT on a blog?
The chatGPT response made more sense. The "pros" are that the system is easy to use. Since ChatGPT has access to a huge web crawl it and write articles on a wide variety of topics and is fun to use.
The con section is well written as well. I mentions that AI is often inaccurate. This class of AI programs is a language model. The system knows how to summarize text from the web sites that it crawled. The system does not actually understand things. The program knows how words are used but does not have direct involvement in interacting with the world.
AI is simply summarizing the work of others.
The second listed con is the "Lack of Authoritative Voice."
This second pitfall is the crux of the matter. The problem is that AI writes clean sentences and a small number of people are using this combined with the buzzword "artificial intelligence" to present text from AI as authorative.
One of the biggest dangers that we face is that AI has an authoritative voice but does not have actual experience. It is likely to lead people astray.
I am glad that ChatGPT introduced the ethical problem. Already forums of the world are being filled with mindless AI generated content.
Overall, republishing a chat with ChatGPT on a blog can be an interesting and time-saving option for content creation. However, it's crucial to carefully review the generated content for accuracy and coherence and to consider the preferences and expectations of your audience when deciding whether to republish AI-generated conversations.
I will end the post with my conclusion.
AI is a language model. AI programs are able to analyze and generate human text.
Since the text is not the product of direct experience or of carefully thought through research, AI is likely to increase the amount of whitenoise on the internet.
I can see some value in using AI as part of a research project as AI is able to scan through reems of information and present persectives that an author had not consider before writing a post.
There are pros and cons to all technologies. AI is likely to improve the quality of writing. It might also do little more than create white noise and re-inforce bad information.
I am happy to see that ChatGPT brought up the ethical issues.
Rather than condemning the new technlogy. I think that web publishers are wise to experiment with it carefully. If you are using AI, then you should make a recording of the conversation as generated and include the AI in a citation so that readers know the source of the information.