Are ChatGPT and Midjourney the only options available? Here’s an article comparing the most famous artificial intelligences with Google Bard and Adobe Firefly. Which one will emerge as the winner?

Scenario: an abandoned rocky and desert field, void of life for decades, bearing the scars of years gone by. Drones hover in the sky, ready to broadcast an event that will change the planet’s fate.

On opposite horizons, flickers of sunlight reflect off shining objects. Suddenly, a rhythmic crescendo of heavy steps approaches. A drone kickstarts the show by focusing on an OpenRB7.7-V14. Counterbalancing this is the emblem of the B.A.R.D 5.7GG attack forces.

The two private armies arrive within 100 meters of each other. Silence falls over the battlefield, but the excitement builds in the stadium-cities across the northern hemisphere.

The first war between artificial intelligences begins.

Back to Today

Influenced by years of consuming sci-fi movies, books, and TV series, this scene evokes a mix of fear and excitement I didn’t expect to feel writing about my experience with some AIs.

Personal interest and fulfilling a childhood dream, kept hidden for years, led me to test some AI options as soon as possible. ChatGPT and Google Bard for writing, Midjourney and Adobe Firefly for image creation.

From this experience, I’ve drawn some comparisons that might help those still deciding which AI to explore.

ChatGPT, Also Known as the Trailblazer

ChatGPT’s popularity has garnered global attention, opening the doors to AI for the general public. Its rapid impact caught governments off guard, leading Italy to temporarily ban its use in March 2023 for data protection reasons (but quickly making it accessible again).

It’s a generative chatbot based on artificial intelligence created by OpenAI. It’s a pre-trained AI that uses Natural Language Processing (NLP). Its data sources are textbooks, websites, and various articles, which it uses to model its language and respond to human interaction via chat input.

The most used version is the GPT-3 model (currently 3.5), while GPT-4 is available in a limited way for Plus users, a paid service that provides general access even during peak periods.

Google Bard and What Happened Yesterday

Bard is also a generative AI. Given the timing, it might seem hastily launched to compete with ChatGPT 4’s release. However, Google laid the groundwork for ChatGPT and other LLMs by making its deep learning model, Transformer, public in 2017. Bard utilizes the Google Language Model for Dialogue Applications (LaMDA), released in 2021.

Thus, OpenAI’s new tool shares a common lineage with Google’s, but Bard has been in development for years. Although not yet available in Italy, a Chrome VPN extension allows access and experimentation with its capabilities.

ChatGPT vs. Google Bard: Comparison Points

Reference Database: Bard directly accesses the internet and Google’s resources, while ChatGPT uses an internal database at OpenAI. ChatGPT’s archive was last updated in September 2021, whereas Bard can provide real-time information thanks to its connection to the search engine. Both remain valid tools for linguistic research and content creation.

Image Integration: Bard can provide answers in image form, being directly connected to the search engine. ChatGPT only provides textual results.

Language: ChatGPT can converse in over 90 languages, including Spanish, French, Arabic, Mandarin, Italian, Japanese, and Korean, but proficiency varies by language, with English being the main one. Bard communicates only in English, Korean, and Japanese.

Drafts: Bard produces multiple versions of its response, allowing users to select the best from the first generation. ChatGPT provides a single, revisable response if it doesn’t meet user expectations.

Price: Both chatbots are freely accessible. ChatGPT requires registration and offers a Plus version providing the latest AI version and priority in requests.

So far, these AIs have been applied in text processing. But how do they fare in image creation?

Midjourney and a Child’s Dream

Midjourney was founded by David Holz, also the co-founder of Leap Motion, a well-known startup in virtual and augmented reality acquired by UltraHaptics in 2019.

Midjourney’s text-to-image AI development started with a closed beta, which opened to the public in July 2022. This allowed public access and use, contributing to its spread and popularity.

The Respectful Adobe Firefly

Many top AI art generators, like Midjourney, Stable Diffusion, and DALL-E2, have been criticized for “scraping” the internet for reference images, regardless of ownership or copyright.

Adobe Firefly, unlike AI models relying on vast centralized data and computing power, adopts a decentralized approach. It relies on Adobe’s resource library as a source for images. Like Midjourney, it

allows for high-quality image creation, but only from images that can be safely used for commercial purposes from Adobe Stock.

Midjourney vs. Firefly: Comparison Points

Training Data: Midjourney is trained on an extensive dataset of images, ignoring copyright. This makes the images aesthetically appealing due to a combination of countless styles and compositions. Firefly, on the other hand, is trained using a publicly available image dataset. Its products are satisfying but may not be as innovative and often resemble stock images.

Prices: With the introduction of version 5, Midjourney ended its free trial period. It now offers three different monthly or annual subscriptions (basic, standard, pro) that vary in cost (excluding VAT), features like generation time available, and the number of simultaneous requests.

Firefly is available for free in a beta version accessible by email registration.

Functionality: Midjourney has a broader set of features compared to Firefly. For instance, it can generate, modify, upscale, create new versions, and derive prompts from images. Firefly users can only create images from textual instructions from scratch, filling spaces, and applying styles and textures to texts.

There are many other applications available, and most software and apps, from Photoshop to Canva, are implementing AI in their platforms. Starting to study and familiarize oneself with these tools is the first step towards opening a door to new perspectives, regardless of our sector.

If we had to choose only one of these tools in a near or distant future, do you already know which side you’d take?