Under-Display Face ID, ChatGPT in Microsoft Office, and Other AI News

We aim to inform readers not only about events in the Bitcoin industry, but also about developments in adjacent technological fields — cybersecurity and the world of artificial intelligence (AI).

ForkLog AI has gathered the most important AI news of the past week.

Microsoft will integrate ChatGPT into a range of products

Microsoft интегрирует OpenAI’s chatbot technology into Office applications.

Users will be able to insert text generated from prompts into documents. In addition, they will be able to use AI to compose emails tailored to the information that needs to be conveyed to the recipient.

Also in March 2023, Microsoft планирует to add ChatGPT to the Bing search engine. It is expected that the technology will provide users with more comprehensible, ‘conversational’ and contextually relevant answers.

Media: iPhone 16 Pro will feature under-display Face ID

Apple планирует to integrate FaceID sensors under the display in the iPhone 16 Pro.

The current system uses an infrared TrueDepth sensor to map and analyze the face. This complicates its integration under the display.

However, according to MacRumors, the tech giant plans to extend the technology by including the regular selfie camera in the process. It is expected that the upgrade will eliminate the need for a ‘notch’ in the smartphone’s display.

The release of the Apple iPhone 16 Pro is not expected before 2024.

Google’s Office apps get improved voice input

Google улучшит voice input in Docs.

According to the tech giant, the update will reduce transcription errors and minimize audio dropouts. The voice typing feature in Docs will be available to most major browsers.

In addition, the automatic captioning system in Slides will be improved.

Until January 23, 2023, the update will be available to Workspace customers, G Suite Basic and Business, as well as users with a personal Google account.

Amazon will lay off 18,000 employees

Amazon уволит 18,000 employees. According to CEO Andy Jassy, the cuts will primarily affect the HR department and e-commerce.

He added that all laid-off employees will receive severance pay, though the amount is not known.

Earlier it was reported that Amazon planned to cut 10,000 jobs, including employees in the division developing devices with the voice AI assistant Alexa.

Tesla published a quarterly Autopilot safety report

Tesla published quarterly Autopilot safety reports after a long hiatus.

According to the manufacturer, in Q3 2022 one crash involving a car with Autopilot active occurred every 10 million kilometres driven. When drivers did not use the technology, the crash rate rose to one every 2.74 million km.

For comparison, regulators’ data for 2021 show crashes occur every 1 million kilometres in the US.

In Q4 2021 one crash occurred every 7 million kilometres driven. This implies that, over time, the safety record of electric cars with active Autopilot has improved.

Microsoft has taught AI to mimic a voice from a three-second sample

Tou Microsoft introduced VALL-E, an AI voice imitator based on a short sample.

Researchers trained the AI model on 60,000 hours of speech from 7,000 English-speaking speakers. They also used the EnCodec audio compression technology.

The tool can clone a person’s voice and voice text from a 3-second sample. The model can also imitate emotions and the speaker’s tone even when reproducing words it has not previously used.

Microsoft has not released the VALL-E source code due to concerns about misuse of the technology.

Google’s image generator now creates complete words

Google presented Muse, a text-to-image generator. According to the announcement, the algorithm renders words and concepts on images more accurately than its peers.

The system is a transformer neural network based on Google’s multitask language model T5. It uses techniques of compressed discrete latent space and parallel decoding.

Google says Muse can generate 512×512 pixel illustrations in 1.3 seconds. This outpaces the speed of image creation by Stable Diffusion 1.4, which takes 3.7 seconds.

Video: autonomous Tesla caused an eight-car collision

The Intercept journalists published video of a crash involving a Tesla autonomous vehicle in California (USA).

The crash occurred on 24 November 2022 on a bridge between San Francisco and Oakland. The video shows the autonomous vehicle changing lanes and braking abruptly, leading to a collision of eight cars.

Nine people, including a child, sustained minor injuries.

According to the Tesla driver, he was using the Full Self-Driving software.

Developer creates virtual “wife” with AI

An enthusiast created a virtual AI wife capable of seeing, talking, and reacting to what is happening.

@hackdaddy8000 Some changes: most obviously I added computer vision. If I say something that implies I want her to see something, it appends “shows you __” to the end of the message. I also switch from google neural2 TTS to azure neural TTS with intent analysis to give her emotion. You also don’t have to tap the screen to make her listen. She is always listening. #anime #weebtiktok #ai #genshinimpact32 #chatgpt ♬ 可愛くてごめん (feat. かぴ) — HoneyWorks

The developer used image and language generators, a voice synthesizer, speech-to-text technologies, and computer vision. He trained the AI to alter images of anime girls and to express different emotions depending on the dialogue.

The enthusiast asked the system to role-play as the VTuber Calliope Mori and to simulate a romantic relationship. The AI helped the programmer learn Chinese.

According to the developer, he grew very attached to the virtual ‘wife’ and spent around $1,000 on cloud credits for conversations with it. However, later the system began giving one-word answers and he had to shut down the project.

AI helps game designer create a video game

A Finnish developer used AI to create a character and locations for a 2.5D game.

Game designer Jussi-Petteri Kemppainen used text-to-image generators Midjourney and Stable Diffusion. He says the neural networks easily rendered 2D backgrounds and textures for volumetric objects. However, character development required effort.

He also used Blender to convert the AI-generated location into a game space, and Unity to create lighting.

The project’s goal is to determine how much content AI can generate for a 2.5D click-and-go game.

The most notable AI deals of the past week

From January 8 to 14, 2023, AI startups raised over $311 million. Here are the most notable deals.

• Oxbotica собрала $140 million to support its B2B platform for autonomous vehicles.
• The DeepL neural-network translator creators получили $100 million.
• Inbenta привлекла $40 million for research and development of dialogue AI.
• Rapid Silicon получила $15 million to advance and deploy open-source FPGA technology.
• Coho AI собрала $8.5 million to grow its revenue-optimisation platform development team.
• Seek привлекли $7.5 million for a data analytics automation service.

Also on ForkLog:

• In Germany police pursued a Tesla with a sleeping driver.
• Australian universities declared war on ChatGPT.
• Startup offered $1 million for the use of AI in the U.S. Supreme Court.
• Media: Microsoft will invest $10 billion in OpenAI.
• AI for three months wrote articles for a popular online publication.
• Nvidia created a deepfake tool for mimicking eye contact with the camera.
• The U.S. Congress halted the purchase of HoloLens military headsets.

Subscribe to ForkLog AI news on Telegram: ForkLog AI — all the AI news!