AL JAZEERA INTERVIEW & AI OBSERVATIONS
Den Bento
AI Artist
Creative Director
Public Speaker
Midjourney
OpenAI
Runway AI
Al Jazeera
This frame was created by Runway video generator based on the clip of my participation in Al Jazeera's "The Stream" show.
Since 1998, I’ve been staying updated with creative tech, starting with Adobe products in college while studying Graphic Design.
In late 2022, I began beta testing popular and emerging AI models for copy, graphics, images, and video.
In 2023, I shared my insights as a Creative Director on Al Jazeera's "The Stream", discussing Midjourney and generative AI.
Is AI better at making art than humans?
May 11, 2023
Popular AI image generators are able to produce seemingly endless amounts of stunning visual art in just a matter of seconds. So where does the technology leave professional artists?
Earlier this year, a group of artists and illustrators filed a class action lawsuit that claims the generative AI tools Midjourney, Stable Diffusion and DreamUp violate US copyright law by repurposing artists’ work found on the open web.
AI image generators are trained on datasets made up of billions of images collected online and generally without the artists’ knowledge or approval. Users can then prompt the AI to create new artwork in the style of a specific artist. For many digital artists, the technology represents a threat to their livelihoods. Yet others see AI as a powerful new medium that can be used for artistic expression.
In this episode of The Stream, we’ll look at how artificial intelligence is disrupting creative work and getting people to re-examine how we think about art.
I focused on sharing my concerns about the potential disruption AI could cause for creators. While many were as impressed as I was, those who continued with it now realize that this will never be considered true art or a substitute for ORIGINALITY.
This is another video generated from the background in the interview which is my Samsung Art Fridge. I believe it took a very nice output for what my lines try to achieve which is ever moving images. It was not at all a bad result from the first version of Runway. Actually it impressed me a lot! But it still not there, it has a very inconsistent animation performance.
Reflections on AI and Creative Ventures
My first creation with the public release of Walle from OpenAI amazed me. It captured my essence as a creative director perfectly through a descriptive prompt, and I used it as my Twitter avatar. However, I’ve since closed all my personal and graphic social media accounts. This decision came after discovering that many AI companies had used images without consent, impacting fellow creatives worldwide. It's a sad reality but we need to evolve past it, get through this, informed.
After two years of exploring AI generative models, I've realized that while AI can produce great ideas, it still needs an expert to finalize the work.
I've curated over 200,000 unique images for future projects, but none meet the high-end quality standards required.
In 2023, I launched Tropicalia Lifestyle, a London-based drop-shipping company specializing in organic bespoke scarves of the highest quality.
Unfortunately, after successfully launching the e-commerce site and social media, several Chinese companies quickly started following me in every channel I was putting the designs out.
Given my previous experiences with copycats on social media, I couldn't risk losing a year’s work of transforming AI into usable luxury materials. That project is currently on hold, with several collections ready to go.
I'm hoping to sell the company to someone with the resources to protect their IP and scale it up.
The creative industry faces challenges not just from AI tools but from bad practices that make safeguarding work difficult.
AI is a valuable tool, akin to other transformative technologies we've embraced over time. It's up to legislators and developers to ensure minimal protection for our work shared online.
For now, I only showcase projects in my portfolio that I own, my entrepreneurships, projects that have expired NDAs or that have been or are public to this day. I am also avoiding the usage of social media for my work, and I am only publishing in contra.com because I got tired of plagiarism and scammers.
This choice ensures that my work remains protected until industry practices improve. Despite some bad experiences, the great results and positive experiences have been the norm.
Here are some of my first generations in 2023. These are experiments and research of ideas for my pattern and surface design projects and a possibility to create brand collections for entrepreneurial opportunities.
Although I liked a lot what I created, I wasn't sure about the origin of the images I was getting inspired from. So up to now, summer of 2024, the only viable possibility is to keep training my private models and using AI as part of the process, not the final outcome.
Exploring packaging and stationary design ideas on the styles of my likings.
Based on my first Tropicalia logo, I generated a merged version of it in different styles of designs, patterns and art styles. This compilation worked well!
I was posting my experiment in my now closed Twitter account. I was part of the emerging AI community that shared their experiments. At some point i created a digital business selling some of the patterns I was creating in collections for download. Soon I realized this was not a good idea, due to copyright problems that may surge in the future. I think there was a lot of true information missing when the tech companies opened their models for public and private use. The first results were pretty good on idea terms, but no real artwork existed. Only vectorizing this images could make them viable for use in any design project. The results I achieved in vectorizer.ai were really good and those were the ones I offered for sale.
The AI Apps I Am Presently Using - Summer 2024
In my creative process, I use a variety of tools: pen and paper, ink, acrylics, pens, and traditional methods before even touching Procreate, vectors, photo editors, video editors, AI generators, and recently, AR, VR and XR tools.
In 2024, I switched to AFFINITY for its seamless integration of pixels and vectors in one program, which I own forever—a big plus for intellectual property and creative freedom.
I highly recommend it! It's an absolute substitute for the recently struggling and corrupt Adobe company. Affinity was recently purchased by Canva which actually sucks, because they suck at copyright, very Meta like company. Faking and making it impossible to opt out their trainings and ownership of our creations.
For graphic work, Kittl has been a great discovery of vector and image generation for basic branding and illustration purposes for merch or social media.
I would NEVER USE Canva for any private work, they have the worst practices and you can NEVER DELETE ANYTHING YOU UPLOAD THERE. And if you complain, like I did, you get locked out and never replied to again. They are data mining companies just like Recraft, Vectorizer.ai, and Bigjpg upscaler and so many so many more companies. They have great tools though, if you are not a professional or someone who needs to protect IP. I have used them, no complains on their tech, just their practices. I would only work in these platforms if a client asks me to, but under the consent that their work will never be private or totally owned.
I no longer use Midjourney or Leonardo, for the same reasons and some others that are worse practices than the others, but I recognize their great job with their tech. They are too dangerous because they are attacking personal identity and copyright in every generation. Their problems have a root, and the root is robbery and impunity.
They are just not clear where their training data came from, and I seriously dislike their practices. The same goes for OpenAI. Even though it offers private models while you pay, the results it gives for professional work are not reliable, and I dislike the monopoly they are trying to achieve. I do use their word generators to brainstorm and correct copy.
PERPLEXITY.AI has been the app that has proven most functional for my work due to its great capacity for summarizing and researching the web at the same time, sharing the sources of its results, and generating questions that help a lot when researching free of halucinations. Very kind too.
Claude is great but has the same problem—it's overhyped, in my opinion, but besides that, it's a great alternative to ChatGPT which I avoid at all costs, IT JUST ISN'T CAPABLE TO BE TRUST WORTHY. I do not use DALL-E 3 of OpenAI any longer because their results are outdated, and there is no variety of styles in the ideas in the generative image models, but all about that company is wrong, I wouldn't use them any longer even if they were the best at the job, I still am aware their data is completely corrupt.
For video, I am very familiar with Runway, and though there are many new great generators of video, I am still waiting for them to get good for real use. The outstanding new releases are LensGo, Luma Dream Machine, and Veed.io. They are extremely expensive for experiments, and it takes a lot of money to create something decent—extremely limited.
NoteGPT is another great tool for research, summaries, and transcripts of video and audio files. Very precise and super fast.
CapCut is a great tool but it has serious copyright problems and they are totally open in admitting they train with your work, so they are also not a tool I will use professionally.
For video creation and edits, my top favorite is DaVINCI RESOLVE, which was built originally for colorizing video but it has sound and effects totally integrated in the editing workflow. Doesn't train on my work, its free for what I use and its locally run.
I'll keep updating this section with more insights. As of July 13, 2024, I'm excited to start my journey on this wonderful platform and look forward to awesome, safe, and productive collaborations. It may take me sometime to curate my favorite projects since I started my freelance career in 1999, I'm on it 😉
Thanks for reading this far and I'm at your service!
Den Bento
