Google, Microsoft and OpenAI announced new artificial intelligence (AI) services this week that range from YouTube to Excel. Google opened the wave of announcements on Monday (18) and Microsoft followed suit this Thursday (21).
The strategy of two of the largest companies to redouble investments in AI that began at the beginning of the year should begin to have more impact on the general public starting this week, with the integration of artificial intelligence such as ChatGPT and Bard into everyday products such as emails, text and spreadsheet editors.
In January, Google and Microsoft announced restructurings, which involved accelerating feature launches with artificial intelligence and machine learning and also laying off employees.
Google then launched Bard to compete with ChatGPT, which began to be present on Microsoft platforms — after billion-dollar financing for the AI developer, OpenAI.
In a series of events held this year, the chief executives of Google, Sundar Pichai, and Microsoft, Satya Nadella, gave indications that text and image generating technologies would be central to changing people’s experience with the companies’ products.
Until the announcements at the end of September, little had reached the traditional solutions of the two big techs. Soon, the public will be able to test them.
The week also included news from OpenAI itself, which announced the latest version of its image generating platform, Dall-E 3. In addition to understanding more nuances and details in user requests, the technology will be integrated into the ChatGPT Plus subscription version ( available for R$ 98.45).
Amazon, owned by Jeff Bezos, also followed the trend and incorporated artificial intelligence into its virtual assistant Alexa. The announcement is this Wednesday (20).
See below the main artificial intelligence launches of the week.
BARD EXTENSIONS AND NEWS ON YOUTUBE
The round of announcements began with extensions for Google’s chatbot, Bard, on Monday (18). With this feature, Google’s artificial intelligence receives access to information available in Google Workspace apps (Gmail, Docs, Sheets and Drive), YouTube, Google Maps, Google Hotels and Google Flights.
The tool has been available in Brazil since Tuesday (19), but only works in English. Google does not mention a release date in Portuguese.
When accessing Bard, Google warns that conversations with AI can be accessed by human reviewers. The technology giant, however, guarantees that these people do not have access to personal data accessed by Bard while using the extensions.
The option, in theory, resolves the limitation of chatbots to the information used during the development of the language model. Bard, which previously could not have access to user information, can now create a travel itinerary based on flight dates or hotel reservations available in Google tools.
New York Times artificial intelligence columnist Kevin Roose tested the tool and reports a string of errors: made-up emails, bad travel advice and poor analytical skills. Roose, however, says the technology works well for simple tasks.
Bard extensions, for now, are only available in personal accounts and can be accessed in the platform settings, accessible from the gear-shaped button in the top right corner of the screen. You also need to change your Google account language to English to access the update. The user can choose which Google applications Bard will have access to.
YouTube, managed by the same holding company that controls Google (Alphabet), also announced this Thursday (21) features with artificial intelligence to facilitate video editing. With the new features, the production experience is more similar to that of TikTok.
Following the trend of artificial intelligence launches this year, Microsoft responded to Google’s announcement by releasing an “AI-based copilot” for the Office suite.
At a closed event for the press held this Thursday (21) in New York, the chief executive of the company behind Windows, Satya Nadella, showed how a user could find data about a scheduled flight among messages with help of the tool called “Office AI” — not very different from what extensions for Bard are capable of doing.
Microsoft will make the tool available to the public starting September 26th. By November 1, all corporate customers should have access to the resources, according to the company.
Microsoft’s AI capabilities work from OpenAI GPT-4 AI models (available in the paid version of ChatGPT) and Dall-e, which generates images.
Amazon’s virtual assistant was integrated with a language model based on AI (artificial intelligence), also called Alexa, similar to ChatGPT. The announcement was made at the technology giant’s global launch event held this Wednesday (20) in Arlington, in the United States.
Integration with AI will allow Alexa to understand phrases and contexts to deliver more appropriate responses. Furthermore, the assistant will be able to respond to multiple requests simultaneously — today, Alexa responds to one at a time.
The new features aim to make the experience with smart home devices like Alexa more interactive, according to Amazon’s senior vice president of devices and business, Dave Limp. The technology also speeds up the assistant’s responses by 40%, according to Amazon’s announcement.
The virtual assistant update will be released “in the coming months” (no prediction) and only in the US, gradually.
The language model used in Alexa’s development focuses on traditional uses of the virtual assistant and should deliver a different experience than interacting with Google’s ChatGPT or Bard, which provide everything from travel itineraries to simple computer programs, Limp told the website specialized Verge.
The most recent of the releases came from OpenAI (startup developer of ChatGPT): the third version of the Dall-E image-generating AI. In addition to greater precision in translating the user’s textual commands into illustrations, the update promises to stop violent, adult or hateful content.
In the field of image generation, OpenAI competes with the platforms Stability Difusion (open source) and Midjourney, used to produce fake viral images of the pope in a white inflatable coat.
Unlike Dall-E 2, available on its own platform, Dall-E 3 will be integrated from October to the paid versions of ChatGPT —Plus (available for R$98.45) and Enterprise (for companies).
Even before the controversies with ChatGPT, Dall-E was the target of criticism and lawsuits for allegedly copying the styles of visual artists, whose work was used to train the artificial intelligence model.
To avoid copyright issues in a paid service, OpenAI says it has programmed Dall-E 3 to refuse requests that ask the platform to copy a certain artist’s style (which may also ask to have images taken from the language model).