A Discussion About AI
Greetings. I would like to discuss the topic the emergence of AI within the public sphere, some of the tools available, and some considerations to take into account when posting content derived from AI.
Most people are familiar with digital assistants. The most well-known being the Google Assistant, Apple’s Siri, Amazon’s Alexa and Microsoft’s Cortana. Theses platforms are often used in conjunction with voice commands to carry out common tasks as playing songs, controlling lights within the home, making hands free phone calls, etc. They also provide answers to questions such weather forecast, sports scores, stock prices, and the like. This technology may be considered the fore runner to today’s AI and has been in existence for a number of years.
AI (artificial Intelligence) technology has evolved and the current focus points the the development of Large Language Models (LLMs). The models are trained on datasets based on billions tokens of that allow the models to give human-like responses when prompted. Perhaps the most prominent among the models is Chat-GPT, developed by OpenAI. The responses are based of statistical word prediction and is aided the the use of neural networks. A good primer for generative AI can be found at the following link: A Beginners Guide to Large Language Models
Over the past two years, there seems to have been an astounding amount of activity surrounding generative AI. The main drivers of this growth comes from the likes of OpenAI, Google, Microsoft, and Meta, among others. These companies provide cloud based solutions. There are of course the open source offerings such as Lama, Mistral, Red Pajama, etc. These models are licensed as open source solutions and place great emphasis privacy and data security. These projects also give users the possibility to run open source models locally on their computers, and in a number of instances, on their mobile devices and even offline.
In addition to text based output, newer generative models, as in the case of Stable Diffusion, DALL-E 3, Leonardo AI, etc., can generate images based on text prompts. Other models provide human-like speech generation from text prompts, while other models offer text to music functionality.
The number of tools available is truly astounding if not somewhat overwhelming. Which bring me to my next point. Where does one begin to zero in of the tools that best match the needs of any given project? Some tools are free while others are subscription based. Most service providers offer tiered pricing usually based on three general categories: free plans, professional, and enterprise. Feature sets are commensurate with the given pricing plan. However, pro and enterprise plans can become rather expensive.
Now on to the topic of AI generated content. While AI tools can simplify the creative process, they do come with some trade offs. In some instances, models have been known to hallucinate, that is, they deliver information that is demonstrably false. Although recent updates to various models have greatly mitigated the number of instances where hallucination occurs, the problem can nevertheless still can arise.
There is also the risk of plagiarism or copyright violations. The models are trained of sets of existing data of which the models may sometimes reproduce verbatim. It goes without saying, publishing of such content could run the risk of legal action and in many cases, lead to being banned from social medial platforms. Therefore, one should take extra care in reviewing the content before posting.
On a related note, many social media platforms place great emphasis on original content created by its members. Purely AI generated content would likely leave the audience with the impression of laziness on the author’s behalf and represent inferior quality.
To avoid the aforementioned problems, it is highly advisable to take a practical approach in the use of AI tools. For example, one could use AI in such things as in outline creation, generating topics for discussion, and proof reading. The content itself should be result of the Author’s genuine effort. And, just like our mussels, creative skills will tend to atrophy with disuse. Therefore, AI should be used as an aid in the creative process rather than a magic wand.