Yesterday 14 March was an eventful day for the tech world. Google announced the expansion of its AI integration with new AI-powered features for Google Workspace such as Google Docs, Gmail, and Slides. Meanwhile, OpenAI scaled up and introduced GPT-4, its latest milestone.
It can be difficult to keep up with the ever-changing space and developments. Here is all you need to know about these two updates.
Also read: Why QR code menus need to die
Google’s AI expansion
Google has introduced PaLM API for developers and businesses to experiment with large language models easily and safely. Along with this, Google is releasing MakerSuite, a tool that developers can use for prototyping. These tools will be made available through a Private Preview, according to Google’s blog. Developers will be able to access PaLM through Google Cloud.
More than three billion people benefit from AI-powered features such as Smart Compose in Gmail or auto-generated summaries in Google Docs on Google Workplace, according to Google's blog. For those who want to use AI to aid their writing process, Google is now introducing new features that will make it easier and will be tested by limited users. Now, in Gmail and Google Docs, people can write down the title of a topic they want to write about, and the content will be instantly generated, much like ChatGPT. For instance, if you want to write a letter asking for leave, Workspace can write that for you, which you can adjust for tone and make edits—making the process faster.
“We’re so excited by the potential of generative AI, and the opportunities it will unlock — from helping people express themselves creatively, to helping developers build brand new types of applications, to transforming how businesses and governments engage their customers and constituents,” Thomas Kurian, CEO, Google Cloud said in the statement.
About four months after the release of ChatGPT, OpenAI has announced its successor, GPT-4, its latest attempt to scale up deep learning. GPT4 stands for “generative pre-trained transformer 4" and is the fourth version of OpenAI. According to the official statement on their website, it exhibits human-level performance in various professional and academic contexts and is “more creative and collaborative than ever before” and “can solve difficult problems with greater accuracy.”
“We’ve spent 6 months iteratively aligning GPT-4 using lessons from our adversarial testing program as well as ChatGPT, resulting in our best-ever results (though far from perfect) on factuality, steerability, and refusing to go outside of guardrail,” the company said in a research blog post.
The widely popular ChatGPT is based on the GPT 3.5 technology built on a deep learning method that uses an artificial neural network to generate text similar to humans. OpenAI states that the difference between GPT 3.5 and GPT 4 is subtle and the difference can be seen when “the complexity of the task reaches a sufficient threshold.” The newest addition is more reliable, creative and can handle complex instructions. Furthermore, GPT-4 considerably outperforms the existing large language model and most state-of-the-art (SOTA) models, according to the statement.
There are three main changes. Images can be fed into GPT-4 to generate captions, classifications, and analyses. The newer version also comes with more creative abilities that can be used for tasks, such as composing songs, writing screenplays, or learning a user’s writing style. GPT-4 has advanced reasoning and instruction-following capabilities. Furthermore, it can handle 25,000 words of text, “allowing for use cases like long-form content creation, extended conversations, and document search and analysis.”
OpenAI also states that it has taken note of all the feedback to improve GPT-4’s behaviour. To improve AI safety and security, they have also “worked with over 50 experts for early feedback in domains including AI safety and security.”
Some organisations that have used GPT-4 include Duolingo, Be My Eyes, and Khan Academy.
Duolingo is using GPT-4 to scale two features: Role Play, an AI conversation partner and Explain my Answer, which explains the rules when you make a mistake. These features will be part of its new subscription tier called Duolingo Max.
“We wanted AI-powered features that were deeply integrated into the app and leveraged the gamified aspect of Duolingo that our learners love.” principal product manager Edwin Bodge said in an official statement on OpenAI’s website.
Be My Eyes connects people who are blind or have low vision with volunteers to help with daily tasks. Now, they are developing a GPT-4-powered Virtual Volunteer within their to “generate the same level of context and understanding as a human volunteer.”
This has the potential to increase global accessibility, said Michael Buckley, CEO of Be My Eyes in a statement on OpenAI’s website. “In the not so distant future, the blind and low vision community will utilize these tools not only for a host of visual interpretation needs but also to have a greater degree of independence in their lives.”
Khan Academy, a non-profit aimed at providing free education across the world, will be using GPT-4 to power Khanmigo, an AI-powered assistant that will function as a virtual tutor for students and a classroom assistant for teachers, according to the statement.
“We think GPT-4 is opening up new frontiers in education. A lot of people have dreamed about this kind of technology for a long time. It’s transformative and we plan to proceed responsibly with testing to explore if it can be used effectively for learning and teaching,” explained Kristen DiCerbo, Chief Learning Officer at Khan Academy.
How can you access GPT-4?
You will need a paid subscription to ChatGPT+, the premium version of ChatGPT. However, if you are a Bing user then the good news is it turns out you are already using it, as Jordi Ribas, Corporate Vice President, Bing and Microsoft has confirmed that GPT-4 is powering the chatbot interactions.