The major artificial intelligence providers in 2023
In 2023, the market for artificial intelligence services is vast, with numerous providers. However, it is important to distinguish between those who are simply resellers and those who have actually created and manage the models.
To identify a reseller, simply check the terms of service and verify that they do not use third-party AI platforms like those mentioned in this article. In that case, it would be evident that they are simply intermediaries. The platform is legally obligated to disclose the use of third-party servers, as your data would be stored on them.
Furthermore, it is extremely rare for a relatively unknown platform to possess AI hardware and models, as the investment required to offer such services is extremely high. Therefore, only the major players in the industry can afford to do so.
Generally, each company focuses on providing services for specific functionalities, including:
- Text Content Writing
- Image Generation and Editing
- Audio Recognition
The most well-known companies are:
OpenAI
OpenAI was founded in 2015 with an impressive capital of 1 billion dollars. The project involved not only Elon Musk of Tesla and Sam Altman, but also other prominent figures from Silicon Valley, such as Peter Thiel, Reid Hoffman, and Greg Brockman, the founder of Y Combinator Jessica Livingston, the Indian IT multinational Infosys, and computer scientists Ilya Sutskever and Wojciech Zaremba.
The initial goal of OpenAI was to create a non-profit organization that would share its research and patents to collaborate with others in the industry. Its innovative approach quickly captured people's attention, becoming a reference point for technology enthusiasts and researchers.
In January 2023, Forbes reported that Microsoft was in discussions to invest an additional 10 billion dollars in OpenAI, valuing it at 29 billion dollars. It is no longer a non-profit organization.
The models provided by OpenAI include:
The system is completely closed source and can be used through API calls or by subscribing to Chat-GPT. Currently (2023), the GPT model is the best for generating text content.
Stable Diffusion
Stable Diffusion was publicly introduced by its development team, Stability AI, on August 22, 2022.
It does not have major companies behind it and differentiates itself in the market thanks to its open-source availability, which makes it completely free (although suitable hardware is still required).
Stable Diffusion is a deep learning model with various functionalities in image processing.
It is mainly used to generate detailed images conditioned by textual descriptions, but it can also be applied to other tasks such as inpainting, outpainting, and generating image-to-image translations guided by a textual prompt.
Here is an example of what it can do:
(Example taken from the official Stable Diffusion Reddit)
MidJourney
MidJourney is one of the many AI image generators that have emerged recently. Unlike its competitors, it offers a more dreamlike artistic style to your requests.
It will likely attract those working in science fiction literature or art that requires a more gothic atmosphere. While other AI generators tend to focus more on photos, MidJourney is more of an artistic tool.
As the website states, it aims to "explore new means of thought and expand the imaginative powers of the human species".
Here is an example of a photo generated in a more fairy-tale style:
MidJourney 5
Pay attention to the latest version, MidJourney 5, as it has incredible improvements in photorealism!
Here is an example of a boy wearing a cap:
(Examples taken from the official MidJourney Reddit)
LLaMA
LLaMA is the OpenSource alternative proposed by Meta to compete with OpenAI's GPT models.
According to Meta, LLaMA will allow other members of the scientific community to study these models in a more accessible and democratic way, even for those who do not have large amounts of infrastructure.
This field is constantly evolving, and LLaMA represents a high-performing and compact solution that promises to make a difference.
The company claims to train its models on trillions of tokens, demonstrating that cutting-edge models can be achieved using public datasets without relying on proprietary and inaccessible data sets.
That being said, it was released on February 24, 2023, and is inevitably lagging behind OpenAI.
(Click here to read our complete guide on artificial intelligence for businesses)