Robo-taxis in Los Angeles, a government chatbot and other AI news

We aim to inform readers not only about developments in the bitcoin industry but also about what is happening in related technology sectors — cybersecurity and the world of artificial intelligence (AI).

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

American man spent nine days in jail due to facial-recognition error

spent nine days in jail due to a facial-recognition system error.

In spring 2022, the man was accused of assaulting a bus driver and stealing his smartphone near Baltimore. An intelligence analyst using facial-recognition software flagged possible matches with a person seen by surveillance cameras. A police officer confirmed.

However, according to Soyer’s wife, they were sleeping at home at the time of the alleged crime. She stated that her husband was taller than the person seen in the footage and differed in age, facial hair, a gap between his teeth, and a twist of his right foot while walking.

Later, the victim pointed to another person — Deon Ballard. He turned out to be 18 cm shorter and more than 20 years younger than Soyer. Ballard’s mother and the arresting officer confirmed the identification.

Maryland law enforcement did not comment on the incident. Baltimore County’s deputy prosecutor John Cox declined to confirm that Ballard and Soyer were arrested for the same crime.

Voice deepfake ‘hacked’ phone banking

Motherboard journalist Joseph Cox duped the bank’s voice ID using a free AI speech-synthesis service.

The reporter used ElevenLabs’ software. He recorded about five minutes of speech, reading aloud sections of the European data protection regulation.

Cox called Lloyds Bank, which uses Voice ID to identify customers by voice. He manually answered the robo-operator’s questions, then voiced them using ElevenLabs’ algorithm. As a result, the reporter was able to bypass the institution’s software and learn the account balance.

According to Cox, he tested several voice AI services, but they had issues or limitations recreating a British accent. ElevenLabs’ system performed well in this regard, the journalist added.

Waymo launches robotaxi services in Los Angeles

The Waymo company launched autonomous-vehicle testing in Los Angeles.

Following a rigorous cycle of validation and safety readiness evaluation, @Waymo is starting fully-autonomous (no human driver) testing in LA. Thrilled by the data confirming, once again, how well our ML-based 5th-gen Driver generalizes across cities! pic.twitter.com/hd0XU5zecT

— Dmitri Dolgov (@dmitri_dolgov) February 27, 2023

For now, the vehicles carry only company employees. The robotaxis will operate in Santa Monica outside traditional rush hours.

Also, the company said the self-driving cars have logged more than 1.6 million kilometres on public roads without a safety driver at the wheel. During this period there were two collisions and 18 ‘minor incidents’.

YouTube announces new AI tools for content creators

In the coming months, YouTube will unveil generative AI tools for content creators.

According to new head of the service Neal Mohan, engineers are developing new features with ‘carefully considered constraints’. With generative AI systems, creators can expand storytelling in videos and raise their productive value, he added.

YouTube will also implement safeguards to ensure responsible use of AI.

Nvidia released a DLSS-like technology for browser video

Nvidia released a new GeForce Game Ready driver with support for the RTX Video Super Resolution AI technology to improve video quality in browsers.

The feature works only with GeForce RTX 40 and 30-series GPUs in Microsoft Edge and Google Chrome. It sharpens edges and reduces artifacts for any web video.

The technology supports video resolutions from 360p to 1440p at up to 144 Hz, and can upscale to 4K. Previously this capability was available only to Shield TV users.

Bing chat gains mood settings

Microsoft added to Bing chat the ability to change the conversation style.

Now the search engine offers three modes: Creative, Balanced and Strict. The first allows the bot to give more expansive, ‘original and creative’ responses, the latter — short, relevant, and fact-based.

Microsoft set Balanced as default. It provides a balance between accuracy and creativity.

The company also increased daily queries in Bing chat to 100 turns.

We have raised the daily cap for Bing Preview users to 100 turns based on your feedback. The limit of 6 turns per session remains in place for now, but we will continue to expand as time goes on. More announcements coming soon!

— Jordi Ribas (@JordiRib1) February 24, 2023

Xiaomi unveils AR headset

Xiaomi unveiled the Wireless AR Glass Discovery Edition, a wireless lightweight augmented-reality headset.

The device is equipped with a Snapdragon XR2 Gen 1 chip, also used in the Meta Quest Pro. Weighing 126 grams, the AR glasses are made from a magnesium-titanium alloy and carbon fiber.

The headset features two microOLED displays with a peak brightness of 1200 nits, connected to a pair of freeform light-guiding prisms. It also has electrochromic lenses capable of adapting to varying lighting conditions.

According to the company, display resolution reaches 58 pixels per degree. This nearly matches the resolution of the human retina, which stops perceiving grain at around 60 pixels per degree.

Axel Springer to replace journalists with AI

The German media group Axel Springer replaced Bild and Die Welt staff with artificial intelligence.

The company took this step to boost publication incomes. According to head of the media group Mathias Döpfner, the layoffs will occur because automation and AI are making many jobs redundant. The technology could improve independent journalism or replace it, the executive said.

Döpfner also added that AI tools such as ChatGPT promise a ‘revolution’ in information and are capable of aggregating data more efficiently than people. He said the layoffs would not affect creators of original content.

There is no data on the scale of the cuts. However, Axel Springer promised not to lay off reporters, article authors and specialist editors.

AI will help Romania’s Prime Minister ‘guess’ citizens’ wishes

Prime Minister Nicolae Ciucă presented the ‘first of its kind’ government AI adviser Ion.

The locally developed system with a voice interface can scan social media and inform authorities about citizens’ proposals and wishes in real time.

According to Ciucă, using AI will be ‘not optional but mandatory’ for making more informed decisions.

Study: some companies replace workers with the ChatGPT chatbot

Resumebuilder, a job-recommendation platform, found that some companies are replacing workers with the ChatGPT chatbot.

In a survey of 1,000 employers, about a quarter said that OpenAI’s algorithm had replaced staff in a range of tasks.

According to Resumebuilder, 66% use ChatGPT for coding, 58% for copywriting and content creation, 57% in customer-support services, and 52% for drafting meeting summaries and other documents.

In recruiting, 77% of companies use the chatbot to draft job descriptions, 66% create interview invitations, and 65% respond to applicants’ inquiries.

According to the report, most employers are impressed with ChatGPT. 55% of executives rate the algorithm’s performance as ‘excellent,’ and another 34% as ‘very good.’

The week’s most significant AI deals

From February 26 to March 4, 2023, AI startups raised more than $150 million. Here are the most interesting deals.

• Typeface received $65 million to develop AI tools for creating marketing texts and images.
• Sesamm secured $37 million to scale NLP algorithms for extracting information from digital content.
• Voicemod raised $14.5 million to expand voice-changing and synthesis technology.
• Qwak raised $12 million.
• Robin AI raised $10.5 million to scale an AI contract-drafting editor.
• Indent raised $8.1 million to continue developing its AI-powered video-review tool.
• Inarix raised $3.3 million to scale a crop-quality analysis platform.

Also on ForkLog:

• Chatbot Bard criticized mass layoffs at Google.
• Meta will create a team to develop generative AI.
• The U.S. Army will deploy facial recognition in drones.
• Media: Elon Musk is assembling a team to create a competitor to ChatGPT.
• Microsoft has integrated the ‘new Bing’ into Windows 11.
• Amazon accused of excessive surveillance of couriers.
• OpenAI opened an API for ChatGPT and Whisper.
• In the US Congress, a bill to block TikTok was introduced.
• Media: Neuralink was denied testing neural chips on humans.
• CNET laid off 10% of staff amid an AI-stories scandal.