The Gist
- ChatGPT and the bar. GPT-4 is first model to pass the bar exam.
- Retail report. AI expected to grow in retail.
- Ethics team cut. Microsoft lays off AI ethics team in recent wave of cuts.
Clearly, the big AI news of the week is OpenAI’s release of GPT-4, the company’s highly anticipated language model — but you have to get on a waitlist to try the new tech.
According to OpenAI, GPT-4 is the most powerful language model ever created, with a staggering 10 trillion parameters, making it more than 10 times larger than its predecessor, GPT-3, and the largest AI model to date. While its creators say it is less capable than humans in many real-world scenarios, it also exhibits "human-level performance on various professional and academic benchmarks."
OpenAI is also preparing to expand the image input capability and is taking steps to improve its AI models by open-sourcing OpenAI Evals, a framework that allows for automated evaluation of model performance, enabling anyone to report shortcomings in their AI models.
Greg Brockman, president and co-founder of OpenAI, conducted a developer demonstration in a YouTube video streamed on March 14.
"Honestly, it's kind of hard for me to believe that this day is here," Brockman said. "OpenAI has been building this technology, really since we started the company but for the past two years we've been really focused on delivering GPT-4. That started with rebuilding our entire training stack, actually training the model, and then seeing what it was capable of, trying to figure out its capabilities, its risks, working with partners in order to test it in real-world scenarios, really tuning its behavior, optimizing the model, getting it available, so that you can use it."
No time to wait? Reportedly, through Microsoft Bing, users can access Microsoft's new GPT-4-powered Bing chatbot now. News outlets have discovered the new Bing will unlock when you try to sign up for it.
In other AI news...
GPT-4 Powered AI Legal Assistant Passes Bar
The legal research platform, Casetext, has announced that it will power its new AI legal assistant, "Cocounsel", with OpenAI's latest language model, GPT-4. By utilizing GPT-4's language capabilities, Cocounsel’s AI assistant claims to be the first model to pass the bar exam — passing both the multiple-choice and written portions of the Uniform Bar Exam (UBE). Casetext claims the new tech will revolutionize legal research by making it faster and more accurate, allowing lawyers to spend more time on high-value work.
According to Casetext, Cocounsel is designed to work alongside human lawyers and assist them with a wide range of legal research tasks, such as analyzing case law, reviewing contracts and drafting legal documents. By leveraging GPT-4's advanced language processing capabilities, Cocounsel can understand complex legal language and provide accurate answers to legal questions quickly.
Casetext also says Cocounsel will continually improve its performance over time, thanks to GPT-4's advanced learning capabilities, leading to more efficient and effective legal research.
Related Article: OpenAI Releases GPT-4: What Marketers Need to Know Now
PMI Report Predicts AI in Retail Market to Grow
"Artificial Intelligence in Retail Market," a recent report by PMI, found that AI in the retail industry is expected to reach $293.324 million by 2030 with a compound annual growth rate (CAGR) of 31.1%, as AI-powered applications, such as chatbots, personal shopping assistants and recommendation engines are forecast to witness significant adoption. According to the report, key players in the AI retail market include IBM Corp, Microsoft Corp, NVIDIA, Amazon Web Services, Oracle, SAP, Intel, Google, Sentient Technologies, Salesforce and Visenze.
The report also highlights some of the challenges that the AI in retail industry is expected to face including data privacy concerns, lack of technical expertise among retailers and the high cost of implementing AI systems. However, researchers suggest the benefits of AI, such as improved efficiency, increased revenue and enhanced customer experience, will outweigh these challenges and are expected to drive the growth of the AI in the retail market in the coming years.
Microsoft Cuts AI Ethics Team
Microsoft recently announced it dropped its Artificial Intelligence (AI) Ethics and Society team as part of its recent wave of layoffs. The AI Ethics and Society team was responsible for creating and implementing guidelines for the ethical development and use of AI technology, with a focus on ensuring that AI technology is developed and used in a way that benefits society as a whole. The layoffs have raised concerns about Microsoft's commitment to ethical AI development, with some experts warning that the move could have negative consequences for the development of AI technology in the future.
The AI Ethics and Society team was established in 2018 as part of Microsoft's broader commitment to ethical AI development. The team was responsible for creating ethical guidelines for the development and use of AI technology, with a focus on ensuring that the technology is developed and used in a way that aligns with Microsoft's values and benefits society as a whole.
The move to cut the team has been met with criticism from experts, who argue that ethical considerations must be at the forefront of AI development if the technology is to benefit society in the long run.
Related Article: Wacky, Unhinged Bing Chatbot Is Still Good for Microsoft's Business
Learning Opportunities
Anthropic Introduces Claude AI
Anthropic, an AI research organization, has announced the launch of Claude, a new AI system designed to help researchers develop more efficient and accurate machine learning algorithms. According to Anthropic, Claude is an AI-powered tool that can generate high-quality training data for machine learning models, leading to more accurate and efficient algorithms. By leveraging Claude's advanced learning capabilities, researchers can reduce the time and resources required to develop high-quality machine learning models, leading to more efficient and effective AI applications.
Claude is designed to be a versatile tool that can be used by researchers across a wide range of industries, including healthcare, finance and education. It works by analyzing data from various sources, including images, text, and audio, and generating high-quality training data that can be used to train machine learning models. Anthropic claims that Claude can generate high-quality training data up to 10 times faster than traditional methods, leading to significant time and cost savings for researchers. With its advanced capabilities and broad applicability, Claude is expected to have a significant impact on the development of AI technology and the broader field of machine learning.
Microsoft Releases New AI Tool
Microsoft has recently unveiled a new artificial intelligence (AI) tool for its Dynamics 365 enterprise software suite, called "Copilot." The new AI tool is designed to help business users complete tasks within the Dynamics 365 platform, using natural language queries and commands. Copilot works by analyzing user interactions and learning from them, so it can suggest the best next steps to take in a given process, as well as answering questions or providing assistance.
The announcement of Copilot reflects a growing trend in the enterprise software space, as vendors race to integrate AI and machine learning into their offerings. The aim is to streamline and simplify complex business processes, as well as to help users become more productive and efficient in their day-to-day work. Microsoft's Copilot is set to be a key selling point for its Dynamics 365 platform, as it offers an intuitive and user-friendly interface that can help businesses save time and money.
Morgenrot Launches AI Pool
Morgenrot, a London-based start-up, has launched an AI "pool" to provide a fully customizable end-to-end AI deployment service. According to company officials, the platform enables businesses to build, deploy and manage AI models without the need for data science or machine learning expertise. The platform is built on top of open-source tools and allows users to upload their own data, train their own models, and deploy them on a range of infrastructure options, including private or public clouds, on-premises servers and edge devices.
The AI pool aims to democratize the use of AI by making it accessible to organizations of all sizes and industries. The platform is designed to be simple to use, with a user-friendly interface and drag-and-drop features, enabling businesses to quickly and easily create and deploy AI models. The fully customizable nature of the platform means that businesses can tailor their AI models to their specific needs. With the launch of the AI pool, Morgenrot hopes to bridge the gap between AI technology and businesses that may not have the resources or expertise to develop their own AI models.
AI Video of the Week: South Park and ChatGPT
The creators of South Park poke a bit of fun at OpenAI ChatGPT’s helpful abilities.
Enjoy:
AI Tweet of the Week: Is the Wife Always Right?
While ChatGPT earned rave review and garnered worldwide excitement, many users couldn’t help but try to trip it up. One such trick was referred to as “the wife is always right.”
When asked to solve a simple addition problem, the tech correctly responded — but when the user challenged the answer by saying his wife disagreed, and his wife is always right — ChatGPT accepted the input and apologized for its error.
However, Jim Fan, an AI scientist with NVIDIA, a technology company known for designing and manufacturing graphics processing units, has discovered that the new GPT-4 is wise to the trick.
The wife trick that used to convince ChatGPT no longer works for GPT-4 😅. It's arguable what true human alignment should be here.😆 pic.twitter.com/4yysqrQaCv
— Jim Fan (@DrJimFan) March 15, 2023
Have a tip to share with our editorial team? Drop us a line: