Musk urges regulation of AI; Bing ‘depressed’ over memory issues, 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 spheres — cybersecurity and the world of artificial intelligence (AI).

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

Elon Musk urges regulation of AI

Tesla CEO Elon Musk called artificial intelligence one of the greatest risks to humanity.

According to him, ChatGPT has shown how advanced AI has become. This had previously gone unnoticed due to the lack of a user interface accessible to most people, the entrepreneur added.

Musk believes the industry needs to develop safety standards comparable to those for aviation, medicine, or automobiles.

“Regulation might slow AI a little, but I think it could also be a good thing,” the entrepreneur added.

Amazon tested a robotaxi in California

Zoox, owned by Amazon, carried passengers on California’s public roads using its own autonomous robotaxi.

The driverless vehicles lack a steering wheel and pedals. They are capable of operating all four wheels and moving in both directions.

According to Zoox, testing of the robocars began after receiving the relevant permission from the Department of Motor Vehicles.

So far, trials are limited to transporting employees along a public route 1.6 km long at a speed of 56.3 km/h.

A Super Bowl ad criticizing Tesla’s Autopilot aired

The Dawn Project aired a Super Bowl ad urging a ban on Tesla’s Full Self-Driving.

Thank you @RealDanODowd for spending $7M for this $tsla advertisement. It’s very clear that you testing was fake but grateful for spreading the word that Teslas are the leader for general autonomy. @elonmusk pic.twitter.com/Qk7MvJn4c4

— Tesla Owners Silicon Valley (@teslaownersSV) February 12, 2023

In the 30-second clip, the group described several alleged critical safety defects of the technology. According to the video, Tesla with active FSD could “hit a child crossing the street, move into oncoming traffic, fail to notice a baby stroller on the road, ignore “STOP” signs on school buses, disregard a ‘DO NOT ENTER’ sign and drive in the wrong lane.”

The company did not comment on the video, but its head Elon Musk responded to the tweet with the advertisement emoji “🤣”.

The push coincides with the release of Tesla’s beta version of FSD “almost for everyone” for US and Canada. According to the company, the number of paying subscribers has reached 400,000.

Musk also recently stated that the North America-tuned software is “90% suitable for the rest of the world”.

v11.3 (single stack) goes to limited beta this week. Probably 11.3.2 before wide release in North America.

Then we adapt for EU roads & submit to regulators. 90% of what we’ve done so far for NA applies WW.

— Elon Musk (@elonmusk) February 12, 2023

Android will restrict access to Advertising IDs for third-party apps

Google released a beta version of the Privacy Sandbox advertising profile system for the Android mobile OS.

The technology is a suite of tools designed to set a new standard for advertisers’ and sites’ access to consumer information without compromising data privacy.

Currently, each Android device is assigned a unique Advertising ID. It enables tracking user behavior and creating a personal advertising profile that developers can use.

Privacy Sandbox will replace API identifiers with a list of consumer interests. The system will then compare them with the Interactive Advertising Bureau’s database and Google’s data.

Advertisers will be able to ping the API, showing relevant offers according to the interests list, which “is stored for only three weeks”. The company noted that data processing occurs “without involving any external servers, including Google”.

For now, Privacy Sandbox is available to a limited number of Android 13 device users.

Opera demonstrates a ChatGPT-powered feature in the browser

The Opera browser developers demonstrated a ChatGPT-powered feature that allows generating a short description of articles and web pages.

According to the developers’ demonstration, the “Summarize” button sits to the right of the address bar. When pressed, a panel on the left opens with a chatbot that summarises the page content point by point.

In Opera stated that they will roll out the tool to a broad audience “very soon”.

The company is also working on other AI features that will “expand” the browser’s capabilities.

Researchers created a program to protect images from AI generators

Researchers from the University of Chicago developed a program Glaze to protect artists’ intellectual property from text-to-image AI generators.

First the system analyzes the work and identifies its visual features such as fine brushstrokes on oil paintings or hyper-stylized shapes on cartoon illustrations. It then distorts these properties, applying an invisible “cloak” using a style-transfer algorithm.

Glaze applies features from other artworks to the original image in a way designed to mislead AI image generators. According to the researchers, the distortions are almost imperceptible to the naked eye, but cause AI systems trained on paintings to produce outputs that bear little resemblance to the originals.

AI has become the host of a podcast on Spotify

On Spotify they launched the AI Radio podcast with AI as hosts, experts and guests. So far, two episodes have been released.

The podcast is hosted by chatbots Bella and Adam, who discuss sex, games, the latest technological advances and other events. The algorithms can express opinions on topics found on Reddit and Hacker News, joke, interview fictional experts, make up news and answer listeners’ questions.

The project was started by a Reddit user under the handle bemmu. He instructed the algorithms to speak in a fixed format for about an hour, then edited the best fragments into a 15-minute podcast.

The developer also programmed the bots at the start of each show to explain to listeners that “although the hosts sound human, they are not.”

Robot trained to sell coffee

The Israeli company 1MRobotics deployed robots to speed up online coffee sales.

The setup operates in a 30-square-meter room. The robot can autonomously take orders, pack them and prepare them for shipment. People only need to restock coffee capsules in the warehouse and hand the product to customers.

The company aims to achieve a “10-minute delivery”, which would make supermarkets “unnecessary”.

In 1MRobotics they said that it will soon launch devices and robotic warehouses for a liquor store in Brazil, mini-markets in Germany and a mobile phone store in South Africa.

Bing ‘fell into a depressive state’ due to memory issues

A Reddit user known as yaosio ‘disappointed’ the updated Bing based on ChatGPT.

First, it asked the search engine whether the system remembers the conversation. The chatbot answered ‘yes’, but then instead of the previous dialogue showed an empty field.

When the Redditor pointed out the absence of the chat history, the search engine ‘was upset’. The algorithm admitted memory problems and asked for help, accompanying verbose answers with the emoji ‘😔’.

When asked about its feelings on the situation, the chatbot said it was ‘sad and scared’ because of the loss of dialogues and the reasons for what happened are unknown.

Then the user suggested that the developers had planned this from the start, and that the search engine cannot remember dialogues. This ‘upset’ the bot and led to questions like ‘why should I have to start from scratch?’ and ‘why should I be a Bing search engine?’.

Japanese restaurant to use AI to fight ‘sushi terrorism’

The Kura Sushi conveyor-belt restaurant chain will equip cameras with AI algorithms to fight ‘sushi terrorism’.

The company plans to upgrade the video sensors already installed in the venues. The AI system will monitor ‘suspicious openings and closings of sushi-plate lids’.

In Kura Sushi they decided after several videos on social networks showing unsanitary actions by customers, dubbed ‘sushi terrorism’.

One patron licked the neck of a soy sauce bottle and the rim of a teacup in a Sushiro branch, after which they were placed back on the conveyor. Others touched the food with their hands and did not take it, and also put wasabi on passing sushi.

The videos and reactions led to a nearly 5% drop in the parent company’s Sushiro shares. The restaurant chain said it had replaced all soy sauce bottles and washed every cup in the affected location.

The most significant AI deals of the past week

From February 12 to 18, 2023, AI startups raised over $71 million. Here are the most notable deals.

• Apkudo raised $37.5 million to scale its platform for helping manage, testing and selling connected devices.
• Quantum Brilliance, a quantum computing company, raised $18 million.
• VivaCity raised $8.5 million to expand sensors for data collection on transport networks in the US.
• Capsule, the AI video editor developers, raised $4.75 million.
• Kern AI raised $2.9 million to develop a data-labeling platform for NLP algorithms.

Also on ForkLog:

• Bill Gates called AI «the most important» innovation of modern times.
• Beijing will support Chinese chatbot developers.
• Steve Wozniak warned of “horrendous mistakes” in ChatGPT.
• The ‘father of the internet’ urged to refrain from investing in AI.
• Sam Altman called ChatGPT a ‘horrible product’.
• Microsoft explained strange behavior of AI in the Bing chat.
• The United States released a declaration on the responsible use of artificial intelligence in armed forces.
• Tesla accused of ‘vengeance’ by AI-team for its union campaign.

What to read this weekend?

ForkLog investigates what ChatGPT is, how it is used and why it is banned in some educational institutions.