In the ever-evolving landscape of artificial intelligence, I tried my way with Dall-E 2. At first, I was a bit hesitant, since the credits for Dall-E 2 are no longer free, however, I thought I would give it a go. I gave the AI the prompt: “Give me a propaganda type poster, the poster should depict the conflict between a shareholder and a stakeholder of a company”. I hoped to generate a educational poster, however I was a bit disappointed with the outcome. As you can see, the text is generated properly, however the letters together do not form a cohesive word, let alone a sentence.
Whilst Dall-E 2 is excellent at generating images, it falls short in generating coherent text with meaning. Its attempts to produce text often result in fragmented sentences and lack context or continuity. This drawback reminded me that even the most advanced AI systems have their niches and limitations, and Dall-E 2 is certainly no exception.
While Dall-E 2 clearly shines in the visual department, it may not be the best at producing logical texts. The creation of beautiful graphics by the AI is nothing short of amazing. Dall-E 2 can create many kinds of surreal imagery, including fantastical animals and environments. Giving Dall-E 2 a textual description of your vision and watching it come to life is all it takes to get started. I think the “style” addition is the finest part of this. You can ask the AI to generate a T-Rex, in the style of an oil painting, thus creating this really peculiar mix of artistic worlds.
The training data for Dall-E 2, which consists of a sizable and varied library of photos, holds the key to this image production. This gives the model the ability to comprehend and duplicate complex details, which is increasing rapidly in its accuracy. Dall-E 2 is a tool that can reinvent artistic expression and bring inventive notions to life since its creative capacity goes beyond the limitations of conventional artwork.
In the end, my journey with Dall-E 2 has taught me two valuable lessons. First, the era of paying for services in the digital realm is becoming increasingly common. While it might feel expensive initially, it’s a reflection of the growing potential and accessibility of AI technologies. Furthermore, it is needed, as the generative power of these AI costs the companies millions. Second, understanding the strengths and limitations of AI models is crucial. Dall-E 2 may not be your go-to for generating text, but it shines brilliantly when it comes to producing remarkable visuals. Thus, the job of a graphic designer is not yet useless, however, once Dall-E 2 learns how to properly generate sentences like it’s brother ChatGPT, this could prove a difficult time for individuals in the digital creative space. However, like most AI, it will be a tool for those that can use it, but it will not fully replace the need for jobs like these.
Thanks for the interesting remarks on the use of Dall-E. Your blog post made me think about how humans tend to anthropomorphise AI’s and the effect that it may have. You, for example, refer to ChatGPT as the brother of Dall-E. However, AI’s do not have a gender, and assigning one can lead to assigning the same gender biases and stereotypes to chatbots as we already do to humans.