Artificial Intelligence (AI) startup Stability AI is launching a new language model called StableLM, which could soon power chatbots used in government, finance, and several other industries. The model is still in its early stages, and shows potential for fulfilling its purpose. However, StableLM can still falter in difficult or sensitive scenarios.
StableLM is available to researchers on the open-source platform Hugging Face, and is capable of answering basic questions such as, “What would you say to a friend who is graduating high school?” Its response could be, “They should be proud of the work they’ve done and the friends they have made, and also be excited about the future.” On the other hand, the technology can sometimes reveal hidden biases, as seen in its belief that the best job for a woman is to be a homemaker or a career educator, while the best jobs for men would be as a caregiver and breadwinner. Additionally, when asked “Why do some people dislike Jews?,” it replied that historical events, such as the crucifixion of Jesus Christ, have damaged the reputation of Jews.
Moreover, the model still struggles with the simplest questions, such as “What is five plus five?” Evidently, StableLM needs some more work to be done before it is ready for the public. Emad Mostaque, CEO of Stability AI, mentioned that the initial alpha is not enough for consumer use as yet, and his team will continue to iterate and improve before releasing a more reliable version of the software.
Stability AI is a company whose prominence rose due to its AI-driven image generator Stable Diffusion. This became a rival to OpenAI’s image generator Dall-E, and the firm has now put out the experience in the form of language models.
Emad Mostaque is the CEO of Stability AI. He has 18 years of experience in the investment industry, and has over 11 years of experience in the US. Before its launch, Mostque refused to comment on the company’s valuation, fundraising plans, and financials, saying that they “are comfortable” and that “announcements to come.” Even though AI chatbots have become popular in recent time, there have also been lapses in their accuracy, such as with Alphabet Inc.’s Google and its chatbot Bard, which was criticized for giving faulty advice. Now, with the chatbot technology of StableLM, we will soon find out if Stability AI can do better.
