AI systems have the potential to revolutionize the digital world, but how will monetization and for-profit systems shape the future of AI?
Open-source vs. For-Profit Artificial Intelligence
Artificial Intelligence (AI) is an umbrella term for a group of computer systems which aim to perform tasks where human intervention is typically required - a good example of high-level AI are self driving cars. The ultimate goal of AI development is an Artificial General Intelligence, which is a type of AI that can achieve almost any general task given to it even better than a human could. This is the type of system often seen in science fiction, typically portrayed as dangerous agents of chaos and evil. Often entire plots revolve around these types of systems obtaining sentience, however in reality it may not be possible for a computer AI to ever become self-aware.
Emerging AI systems
Open AI is a company dedicated to developing AI systems and ensuring their benefits are available to everyone. Their lofty goals include building a safe and open Artificial General Intelligence, the type of system that could help run our lives in future years to come. Not only does Open AI develop these systems, but they also provide API access as well.
Developed by Open AI, GPT-3 (Generative Pre-trained Transformer 3) is an algorithm which can be trained on a set of text, and when given a prompt it can produce new text as a continuation of the prompt. This type of learning system uses the training data to form a model, which is then used to generate the new results. This means that any results can be significantly skewed by the content of the training data.
Open AI provide a playground where you are able to use trained GPT-3 systems online. We gave it the prompt "What are the dangers of AI?" and it produced the following answer; "Some dangers of AI include the potential for machines to become smarter than humans, the possibility of robots taking over jobs, and the threat of AI being used for malicious purposes."
Open AI's generative systems can be used to generate more than just text. In the case of Dall-E, the system has been designed to generate new and unique images based off simple text prompts. It manages to generate some truly astounding results, sometimes nearly indistinguishable from real photos or art.
Left: Dall-E image generated from the prompt "cat with face in chip packet, mood lighting". Right: Dall-E image generated from the prompt "a photo-realistic ground-sloth dressed as Donatello from TMNT".
Copilot and AI Monetization
Copilot is a system designed by GitHub in order to help software developers by suggesting code completions. It uses a natural language model created by Open AI and is trained on hundreds of thousands of publicly available repositories hosted on GitHub. Developers are able to access Copilot to aid in programming for a monthly or annual subscription.
However, GitHub's Copilot has garnered controversy recently; the issue is that despite being trained using open-source information, Copilot is a for-profit product. There seems to be a widespread disregard for the licenses of the individual projects used for training the AI, especially those that include a 'copy left' license. Copy left licenses require that any projects that use them are also made open source. Since Copilot is not open source but is trained using projects that have copy left licenses, Microsoft and GitHub are therefore not honoring these licenses. There has been public outcry from the free and open-source community, including the The Software Freedom Conservancy (SFC) who have migrated away from GitHub and are urging all other Free and Open Source Software (FOSS) projects to do the same.
Through these actions Microsoft and GitHub have shown that they are willing to prioritize profit over the fair and level stewardship of open source projects, setting a concerning precedent as AI systems continue to improve and become more implemented in our day-to-day lives.