According to reports, an AI language model will defend a real-world defendant in a pending court session. The CEO of the startup DoNotPay, which focuses on consumer rights, claims that the objective is to show that AI can take the role of lawyers.
Large language models, like ChatGPT, create text and, after being trained on webpages scraped from the internet, learn to model the relationships between words. Unfortunately, as shown by their training data, their outputs frequently need to be directed away from delivering information that is biassed and discriminatory. They can also produce fake “facts,” which makes them challenging to use in high-risk fields like law or medicine.
Joshua Browder, the CEO of DoNotPay, nevertheless thinks the technology is prepared for widespread use. He claims to have convinced someone to use an AI chatbot to represent themselves. According to reports, the defendant will use headphones to receive audio clips of legal arguments generated by a language model, which they will purportedly recite word by word at a court hearing over Zoom.
According to Politico, the prank will use GPT-J, an open source language model that was made available by the Eleuther collective of AI academics and engineers. Concerned that judges would object to his prank, Browder chose not to provide any additional information about the robot lawyer case that is set to be heard next month.
He has even gone so far as to offer $1 million to any attorney who uses artificial intelligence in a future Supreme Court case, which is challenging given that the majority of electronics are prohibited.
Coming soon is ChatGPT Professional.
OpenAI intends to make money off of the enormously well-liked language model ChatGPT and is determining how much to charge customers for the new product, which is aptly called ChatGPT Professional.
How much should a monthly membership to the model cost? was the subject of a survey that the corporation posted online.
The internet has been overrun by ChatGPT. People like playing around with the tool to programme it to produce various types of content automatically, from essays to jokes. The present system is free and accessible to anybody to use, although it’s often swamped by queries and can be fairly slow in providing results. However, the paid version promises to be quick and constantly accessible.
Although creating accurate text that resembles human speech can seem like innocent fun, there are drawbacks. Language modelling techniques could be abused to promote misinformation and fabricate stories, according to experts. Attempts to ban poisonous hate speech often fail, and tools like ChatGPT often propagate biases in training data.
The technology has also prompted a larger discussion about how it may affect education. Teachers worry that ChatGPT’s sophisticated natural language processing features will make it simpler for students to cheat. For instance, given a text prompt that describes an essay question, language models can produce assignment responses.
Makers of the COVID-19 vaccine purchase Instadeep for £562 million.
InstaDeep, a British AI drug design business, is being acquired by BioNTech, best known for working with Pfizer to create a COVID-19 vaccine, in a deal worth up to £562 million ($682 million).
In a statement released this week, Ugur Sahin, CEO and co-founder of BioNTech, said, “The acquisition of InstaDeep allows us to leverage the fast developing AI capabilities of the digital world into our technologies, research, drug discovery, manufacturing, and deployment processes.
Our goal is to develop BioNTech a technology business where AI is ingrained into every facet of our operations.
If the transaction is successful, BioNTech will advance £362 million in cash and buy out all of InstaDeep’s shares, with the startup’s investors potentially receiving up to £200 million ($244 million), depending on the completion of certain milestones. Both businesses are actively working together to develop mRNA-based therapies in addition to technologies to identify and track SARS-CoV-2 variations.
Drug design with AI is developing. Its promise to change drug creation is driven by new developments in AI model designs and the accessibility of training data. In the hopes of one day curing diseases, new medications, such as antibodies or chemical molecules, are synthesised in labs using algorithms.
Meta and Shutterstock agree to collaborate on AI training
Shutterstock, the industry leader in stock photography, is granting Meta access to millions of pictures, videos, and tracks from its platforms to help the company’s machine learning algorithms grow and be trained.
Training data is essential for neural network performance. To learn from, they require a lot of high-quality examples in big amounts. Shutterstock has the data, while Meta has the tools to create and train models. Since it is more carefully chosen than simply pulling content from the internet, the content is also more likely to be of higher quality.
According to Paul Hennessy, chief executive officer of Shutterstock, “AI has the potential to spark an explosion of creativity.”
In order to further our objective to invest in technology that accelerates the creation of new ideas, Shutterstock is extending our long-standing connection with Meta. This follows our relationships with OpenAI and LG AI Research, which were announced last year.
“In addition to assisting creators in determining how their work is resonating, our goal is to automate the busywork necessary to finish creative endeavours. We will continue to build the future of generative AI in a way that is moral and artist-centered, and our content will remain at the centre of these breakthroughs both on and off our platform.”
