Everything, Everywhere All At Once has just swept the Oscar, winning not one, not two, but seven Academy Awards, including the Best Picture award. . So, let us all stand and applaud this extraordinary achievement, and celebrate the creative brilliance that has made this film a true masterpiece. The winning of EEAAO made cinematic history, becoming the most awarded movie of all time.

Last week, we also witnessed history in the development of AI. Over the past week, the field of artificial intelligence has seen some groundbreaking advancements that will change the way we interact with machines forever.

The release of GPT-4, the latest iteration of OpenAI's groundbreaking language Model, and the release of Copilot, an AI-powered future of Office documents, are among the most significant developments.

These advancements represent a quantum leap forward in the field of AI, and will undoubtedly have far-reaching implications for our world. So, let’s all marvel at the incredible progress that has been made, and imagine the endless possibilities that lie ahead as we continue to push the boundaries of what machines can do.

OpenAI just released GPT-4

Did we just witnessed a revolution?

After months of rumors and speculation, OpenAI has announced GPT-4: the latest in its line of AI language models that power applications like ChatGPT and the new Bing.

The model is "more creative and collaborative than ever before," according to the business, and "can solve tough issues with better precision."

Although it can interpret both text and picture input, it can only answer in text.

GPT-4 can do a lot of great things….

It can process images alongside text

This is the biggest difference between GPT-4 and its predecessor.

The system is multimodal, meaning it can parse both images and text, whereas GPT-3.5 could only process text. This means GPT-4 can analyze the contents of an image and connect that information with a written question. However, it cannot produce answers with pictures yet.

In practice, what does this mean?

The New York Times describes one demonstration in which GPT-4 is shown the inside of a refrigerator and asked what meals may be made with the supplies.

GPT-4 generates a few samples, both savory and sweet, depending on the image.

But, one of these choices — a wrap — calls for a component that doesn't appear to be present: a tortilla.

In a post-announcement demo, OpenAI demonstrated how GPT-4 can generate code for a website based on, for example, a hand-drawn drawing.

This isn't totally novel, but OpenAI believes GPT-4 can "“generate the same level of context and understanding as a human volunteer”"—explaining the environment around the user, summarizing complex webpages, or answering queries about the pictures.

In the future, we can see a lot more of other applications for this functionality.

It’s better at playing with language

According to OpenAI, GPT-4 excels at problems requiring creativity or complex thinking. It's a difficult claim to evaluate, but it appears to be correct based on certain testing we've seen and completed (albeit the differences between it and its predecessors aren't shocking so far).

OpenAI co-founder Greg Brockman requested GPT-4 paraphrase a chunk of a blog article using just words beginning with "g" during a corporate demo. (He later requested that it do the same with "a" and "q.")

More textttttttttttt

The quantity of text that AI language models can maintain in their short-term memory (that is, the text included in both a user's inquiry and the system's answer) has always been a limitation.

However, OpenAI has dramatically extended these possibilities for GPT-4. The system can now analyze whole scientific articles and novels in a single pass, enabling it to answer more complex queries and link more facts in every given inquiry.

It is already employed in popular products

GPT-4 is closer than you might think.

As part of its GPT-4 announcement, OpenAI shared several stories about organizations using the model.

They include Khan Academy's AI tutor tool, which is intended to aid students with assignments and provide teachers with lesson ideas, and an integration with Duolingo, which promises a similar interactive learning experience.

Duolingo’s offering is called Duolingo Max and adds two new features.

One will provide a "short explanation" of why your answer to an exercise was correct or incorrect and allow you to ask for further examples or clarification. The other option is a "roleplay" mode that allows you to practice using a language in various circumstances.

It changed the way we were browsing

With the announcement of OpenAI, Microsoft acknowledged that the model powering Bing's conversation experience is, in fact, GPT-4.

Microsoft's Bing AI-powered chatbot is now available in a sidebar within its Edge browser. The new sidebar shows as a large Bing button in the upper left corner of Edge, and hovering over or clicking on it launches the new Bing chatbot experience.

If you're a Bing preview member, you may use the chat function to ask the chatbot questions or the write option to produce content for emails, blog posts, letters, and other purposes.

Because the sidebar is contextual to the page you're seeing in Edge, you may ask the chatbot to summarize information on a shopping site if you're looking at a product. The Bing sidebar may also summarize articles, compare objects on a webpage, and produce content in blog posts, emails, bulleted lists, and other formats.

This new Bing sidebar in Edge suggests where Microsoft may go with its AI features for Windows, Office, and other platforms.

Well, it can also do your tests for you

One notable metrics in OpenAI's technical report on GPT-4 was its performance on a variety of standardized tests, including BAR, LSAT, GRE, a number of AP modules, etc.

… but it still has limitations

OpenAI also warns that the systems have many of the same flaws as prior language models, such as the ability to make up information (or "hallucinate") and write violent and damaging content.

It will be accessible as an API for developers to build on. Let’s wait for more products built upon GPT-4, or maybe even GPT-5, in the future.

Microsoft Introduces Copilot

What it can do

Microsoft announced a new AI-powered Copilot for its Microsoft 365 applications and services, aiming to help people create documents, emails, and presentations, among other things.

The Copilot, powered by OpenAI's GPT-4, will sit alongside Microsoft 365 apps as an assistant, appearing in the sidebar as a chatbot that Office users can summon to generate text in documents, create PowerPoint presentations based on Word documents, or even assist with features like PivotTables in Excel.

Users of Microsoft 365 can request that Copilot provide information on an upcoming Microsoft Teams meeting, giving users updates on related projects, organizational changes such as new hires, and even updates on employees who have returned from vacation.

Copilot may also be launched from within Microsoft Office products and used in Word to generate documents based on other files. The AI-generated text may then be freely tweaked and updated. Because Copilot is essentially a chatbot, you may even request that it create a 10-slide PowerPoint presentation from a Word document, or that it analyze or prepare Excel data.

This implies that Excel users may utilize Copilot to build an analysis or a PivotTable based on data in an instant. The Copilot function in Microsoft Teams can transcribe conversations, remind you of anything you might have missed if you arrived late, and even summarize action items during a meeting.

Copilot will also be available in Outlook, allowing you to spend less time deleting and replying to emails. Email conversations may be summarized, and Copilot can even generate draft answers using toggles that allow you to change the tone or length of an email.

Microsoft is also planning to launch a Business Chat feature that works across all Microsoft 365 data and apps. It uses the Microsoft Graph to bring together documents, presentations, emails, notes, and contacts into a single chat interface in Microsoft Teams that can generate summaries, planning overviews, and more.

“We make it clear how the system makes decisions by noting limitations, linking to sources, and prompting users to review, fact-check, and adjust content based on subject-matter expertise,” says Spataro, noting the company’s AI principles.

Some concerns

That’s left some inside and outside Microsoft concerned about the pace of AI-powered software, but Spataro doesn’t share the same concerns. “To serve the unmet needs of our customers, we must move quickly and responsibly, learning as we go,” says Spataro. “We’re testing Copilot with a small group of customers to get feedback and improve our models as we scale, and we will expand to more soon.”

Within the artificial intelligence department, Microsoft recently laid off its ethics and society team. The team had been tasked with identifying the hazards presented by Microsoft's use of OpenAI's language models in its products and services.

Some people within and outside Microsoft are concerned about the speed of AI-powered software, but Spataro isn't. “To serve the unmet needs of our customers, we must move quickly and responsibly, learning as we go,” says Spataro. “ We’re testing Copilot with a small group of customers to get feedback and improve our models as we scale, and we will expand to more soon.”