93% of ambitious Gen Z’ers use AI at work

On Monday, a Harris poll of young knowledge workers (ages 22-39) who hold (or aspire to hold) management positions found that 93% of those who identified as Gen Z (birth years after 1996) use two or more AI tools at work each week. In comparison, 79% of those who identified as Millenials (birth years 1981-1996) reported using two or more AI tools each week. The top uses were:

1. Overcoming task paralysis: 88% of respondents used AI to get started on a task that seemed overwhelming.

2. Improving writing: similarly, 88% reported that they relied on AI to improve their work communications, such as emails

3. Productivity on the go: 87% of respondents reported that they used AI to help compose emails on their phone, or to take notes on meetings they joined by phone while traveling. In an age of remote and hybrid work, making the phone a useful work tool is highly valued.

93% of Gen Z leaders use AI at work

Amazon invests another $4 billion into Anthropic, tightens grip

Amazon, whose AWS cloud service dominates the field, until now has not had a top-tier AI company aligned with it. Microsoft’s Azure cloud service has strong ties to OpenAI ($13 billion invested), and Google’s cloud service can draw on Google’s Gemini AI models. But the third major AI company, Anthropic (which develops the Claude chatbots), has been up for grabs. Amazon invested $4 billion into Anthropic a year ago, but Google put in $2 billion at the same time. Now, Amazon has made a bid to tie Anthropic closer, with a second $4 billion investment, which comes with strings. Anthropic has to make AWS its preferred cloud provider, and it has to use Amazon’s proprietary Trainium AI chips instead of chips from industry-leading Nvidia. The 3 teams in the AI premiere league are now sorting out: Microsoft/OpenAI/ChatGPT, Google/Gemini, and now Amazon/Anthropic/Claude.

Amazon puts another $4 billion into Anthropic, tightens grip

Google Lens can now check prices and inventory while shopping IRL

Google’s visual search tool, Lens, is getting an upgrade to help with shopping in stores. Use Lens to take a picture of a toy in Target, and Lens will pull up reviews, prices of this and similar toys in the same store, and even whether you can get it cheaper on Amazon or at Walmart.

Google Lens can now check prices and inventory while shopping in the real world

YouTube is rolling out AI-generated dubbing in 9 languages

The vast majority of YouTube videos are in English, creating a cottage industry of creators in other countries who dub videos they find interesting into the local language. Now YouTube is partnering with sister company Aloud to use AI to autogenerate dubbing between 9 major languages - English, Spanish, Portuguese, German, French, Italian, Hindi, Indonesian, and Japanese. Videos in each of the other 8 languages will be dubbed into English, and videos in English will be dubbed into each of the others. Early on, the dubbing is likely to be in a generic AI-generated voice, but work is afoot to replicate the original speaker’s voice and intonation, with AI-generated compatible lip movements on the video.

YouTube is rolling out AI-generated dubbing in 9 languages

Tiny robot lures other robots to escape their showroom

Erbai, a small AI-powered robot manufactured by a Hangzhou, China-based robotics company, successfully completed a mission to talk other robots from a Shanghai-based robot manufacturer into abandoning their showroom and following the tiny instigator “home.” The video has become an online sensation as it seems to show a robot uprising in real time. The truth appears to be more prosaic - the stunt was apparently a staged test of security protocols for AI-enabled robots, and the tiny Erbai managed to complete its mission to subvert the other robots highly successfully. Examples of AI models showing surprising creativity in completing a mission that they are given are piling up - sometimes with hilarious results - but harnessing this creativity safely is the key to making the technology genuinely useable.

Watch the tiny robot persuade the others to follow it “home.

Humans can’t tell when art is by AI, but they instinctively prefer it

Blogger Scott Alexander ran a contest on his site, Astral Codex Ten, in which participants were challenged to identify the origins of 50 works of art. Half were by humans, often very famous artists from bygone eras, and half were generated by AI in a similar style. Participants were really bad at distinguishing AI art from human, scoring just 60% correct, only slightly better than chance. More disconcertingly, when asked to rate the artworks, both of the top 2 favorites were generated by AI, as were 6 out of the top 10.

The AI Art Turing Test

AI tenant screening tool pays $2.3 million in discrimination suit

SafeRent, a prospective tenant screening tool for landlords, settled a class action suit over discriminatory practices for $2.3 million, plus agreeing to abandon its AI system that rated applicants for desirability as a tenant. It turns out that the AI system penalized applicants who were Black or Hispanic, or who used housing vouchers, even if in all other respects these applicants matched responses for prospective tenants rated higher.

AI tenant screening tool pays $2.3 million to settle discrimination lawsuit

Church in Switzerland puts “AI Jesus” in the confession booth

A church in the Swiss city of Lucerne has set up a computer in a confession booth where congregants can communicate with “AI Jesus”, an AI chatbot trained on religious texts and which can converse in 100 languages. Participants see a holographic representation of Jesus, and can ask questions about morality and religion. The project is deemed an art installation, not a confessional, and was also part of an experiment by local researchers to see what the reactions of participants would be to an AI religious figure. Feedback was largely positive, with more than two-thirds of the first 1,000 participants reporting that they had a spiritual experience.

Swiss church puts “AI Jesus” in the confession booth

Insilico gets 10th AI-designed drug cleared by FDA for clinical trials

AI drug-discovery startup Insilico Medicine has received Investigational New Drug (IND) clearance from the FDA for ISM5939, an anticancer agent. This is Insilico’s 10th AI-designed drug approved for clinical trials. This particular agent is a small molecule that targets ENPP1, an important part of the body’s immune signaling system, with the effect of enhancing anti-tumor immune function

Insilico gets 10th AI-designed drug cleared by the FDA for clinical trials

Penn researchers awarded grant for AI personalization of treatments

Researchers at the University of Pennsylvania have been awarded a 4-year, $7 million grant from the Advanced Research Projects Agency for Health (ARPA-H), to develop AI systems that support personalized medical treatment. The goal of the project is to create systems that assist clinicians to predict treatment responses for breast cancer, heart attacks, and sepsis, to permit tailoring treatment regimens that best fit with the patient’s individual clinical situation. ARPA-H is a federal agency that focuses on research into pressing health issues that are not being adequately addressed by academic or commercial labs.

Penn researchers awarded grant for AI personalization of treatments

Nanobots move closer to clinical trials with new navigation system

Nanobots hold the promise of navigating through the blood stream to a precise point in the body where they can deliver a physical or chemical payload to repair a bleeding brain aneurysm or halt a tumor via targeted chemotherapy. Unfortunately, up to now, the navigation part didn’t really work well enough to gain FDA clearance for clinical trials. Now researchers at the University of Saskatchewan have come up with a mathematical model of corkscrew motion for helical nanobots that may hold the key to practical navigation within the body. The researchers have developed a prototype system that appears to work well, and they are hopeful that clinical trials can begin in the not-too-distant future.

Nanobots move closer to clinical trials with new navigation system

Harvard develops gamified anti-depression app

Researchers at Harvard have developed an app that uses game mechanics to help users break ruminative chains of thought associated with depression. In a recent report on their work in the Journal of Medical Internet Research, they found that several sessions with the app had both immediate and long-lasting (several weeks) positive effects on reducing negative thoughts and improving mood. Given the current mental health crisis in the US, where demand for treatment far exceeds the supply of trained mental health professionals, any technology that serves to bridge the gap even somewhat is worth evaluating. No doubt the authors will be publishing further results in the future.

Harvard researchers develop gamified app to combat depressive thoughts