All the big tech companies are jumping on the artificial intelligence bandwagon, and some foundations are starting to show interest in this field. Mozilla, famous for its Firefox browser, has created an industry-focused startup called Mozilla.ai.
The Mozilla.ai comes with a slightly different mission: to create open-source, “reliable” artificial intelligences. Who says this is Mark Surman, executive chairman of Mozilla and head of the new startup
“Very interesting new technologies are coming along — they immediately took artists, founders… all kinds of people doing new things,” Surman said in an interview with TechCrunch. “Anxiety hits when you realize that almost no one is looking at the limits of it.”
ChatGPT was used to create malware
Many of the technologies released by large companies throughout 2022 have presented problems—and we’re not talking about getting information wrong or anything like that.
ChatGPT, for example, can be used to develop malware, create phishing emails, and find flaws in open source systems. The limitations created by OpenAI were easily circumvented.
Stable Diffusion, which creates images from text requests, was used to create pornographic content, deepfakes and images of explicit violence.
Mozilla has tried, but no one has been interested
The executive recalls that Mozilla has been working with research on reliable artificial intelligences for the public interest for five years. Even so, other companies were not interested in using what their company developed.
“In the middle of last year, we decided we needed to do this on our own, and find like-minded partners to work together,” Surman said.
Mozilla.ai wants to create a foundation for reliable AI
The Mozilla.ai received a $30 million investment from the Mozilla Foundation, which will be its sole owner. In this first moment, the startup has people with passage through Harvard, Huawei and advice of the American government on AI.
Surman plans to build a team of 25 engineers, scientists and product managers. They will work on recommendation systems and large language models (LLMs), along the lines of OpenAI’s GPT-4. The difference is that the focus is on making them in a way that is reliable.
On the horizon, the newly founded company wants to create a network of allied companies and research groups to carry forward its position of developing responsible intelligence.
The idea is to create on a technology that serves as the basis for other products — Surman cites Linux and Apache as examples.