Microsoft is making an upgraded version of OpenAI’s autocomplete software GPT-3. As part of its suite of Azure cloud tools available to business customers. The new generation of AI language models is GPT-3. This system mainly works as an autocomplete tool, and if fed with a snippet of text, whether a poem or an email, the AI will do its best to continue what has been written.
Their ability to parse language also allows them to take on other tasks like generating ideas for projects and stories, analyzing the sentiment of text, summarizing documents. Microsoft says its new Azure OpenAI Service will help customers. To provide customers with the latest upgrades, OpenAI will continue selling its API for GPT-3. Microsoft’s repackaging of the system is meant for larger businesses that want more support and safety. That means the company’s service will offer tools like scaling capacity, data handling protections, private networking, and access management.
Till now, it is not clear how much this might eat OpenAI’s business, but the two companies already have a tight partnership. Microsoft invested $1 billion in OpenAI in 2019 and became its sole cloud provider. Microsoft bought in September 2020 an exclusive license to directly integrate GPT-3 into its products. So far, these efforts have focused on GPT -3’s code generating capacities. Microsoft uses the system to build autocomplete features into its Visual Studio Code editor and its suite of PowerApps applications.
These limited applications make sense, given the enormous problems associated with large AI language models like GPT-3. Microsoft is introducing various safeguards with GPT-3. These include filtering and monitoring tools to help prevent inappropriate outputs or unintended uses of the service, vetting customers’ use cases, granting access to use the tool by invitation only. It is not clear if these restrictions will be enough. According to experts, Microsoft’s reassurances are lacking in substance.
As per the press release, GPT -3’s training data potentially includes –
- Vulgar language
- Racial stereotypes
- Personally-identifying information
Introduction of GPT-3 fails stereotypes fails to meet Microsoft’s own AI ethics guidelines, including a principle of transparency. GPT -3’s training data composition is a mystery. Microsoft claims that the system understands the language. Although Microsoft’s GPT-3 filters are unproven, they can avoid a lot of trouble by simply selecting its customers carefully. As long as humans check significant language outputs, they are helpful. Experts say Azure OpenAI Service is not too problematic if it is just helping to write communication for business executives. Experts are also showing more concern about language generated for a video game character because, without human oversight, this implementation would likely run.
What is GPT-3 –
Generative Pre-trained Transformer 3 is an auto-aggressive language model to produce human-like tests using deep learning. The GPT-n series is the third-generation language prediction model and created by OpenAI. 175 billion machine learning parameters capacity is there in the full version. In May 2020, was introduced. According to the experts, the quality of the text generated is so high that it can be challenging to determine whether or not a human wrote it. It has both benefits and risks.