Neuralink patient plays chess with his mind

Videos of the first patient to receive a brain implant from Elon Musk’s Neuralink brain-chip startup show him in good spirits, playing computer chess with just his thoughts. 29-year-old Noland Arbaugh, who became paralyzed from the shoulders down 8 years ago in a diving accident, volunteered for the implant, and in the video seems fully pleased with its performance. This FDA-approved trial points the way toward greater independence for persons with disabilities that hamper their ability to use tools.

Neuralink patient plays chess with his mind

Stability’s CEO “resigns” after totally screwing the pooch

AI image-generation startup StabilityAI just lost its controversial CEO, Ego Messed-up…erm, I meant, Emad Mostoque. Known largely for his self-aggrandizing appearances on podcasts, Mostoque was by all evidence a terrible CEO. His most senior AI talent were leaving in droves, the company’s flagship product Stable Diffusion was floundering, and investors were getting increasingly restless. Recent efforts to sell the company failed to find a buyer, so in the end, Mostoque “resigned” to spend more time taking selfies.

Stability’s CEO “resigns”

…but takes a selfie with Microsoft CEO (soon his new boss?)

Cheeky to the end, Mostoque released a pic on X/Twitter showing him in a video call with the CEO of Microsoft, Satya Nadella, mere moments after “resigning.” Nadella has been hoovering up scads of AI talent lately (see last week’s story about his acqui-hiring of the entire team at unicorn startup InflectionAI), so the only question is, what’s a smart guy like Nadella doing with a showboat like Mostoque?

Emad Mostaque plays footsie with Microsoft CEO

Google, Intel, and Qualcomm ally to break Nvidia’s dominance

Nvidia is the undisputed leader among AI chipmakers, a status they have ridden to a trillion-dollar market cap. This has dimmed the prospects of AI also-rans such as Google, Intel, and Qualcomm. So the B-team is uniting to attack the true source of Nvidia’s dominance - and it’s not the chips. It’s the software. Nvidia spent years developing GPU firmware, known as CUDA, which efficiently uses the parallel processing of graphic processing units. Google, Intel, and Qualcomm are leading a consortium to develop an open source equivalent to CUDA, to unlock the power of GPUs from any manufacturer. This is exactly the strategy that Google used to develop the android platform for smartphones, to compete with Apple’s proprietary IOS for the iPhone. We’ll see if history repeats itself.

Google, Intel, and Qualcomm unite to break Nvidia’s dominance

Apple partners with China’s Tencent for Vison Pro content

Apple needs content to drive adoption of its Vision Pro “mixed reality” headsets. Unfortunately, Apple has no footprint in China. In order not to be locked out of one of the world’s largest potential markets, Apple has partnered with Tencent, China’s largest online media company.

Apple partners with China’s Tencent for Vision Pro content

Startup Databricks releases most powerful open-source AI

Databricks started out as a data infrastructure company, but pivoted hard to AI a year ago with the acquisition of enterprise-AI startup Mosaic. Apparently 1+1 in this case made the nth power of 10, because their first joint product is a banger. This week Databricks released DBRX (get it?), which instantly became the most powerful open-source LLM in the world. DBRX is outclassing former leaderboard champ Llama 2 from Meta/Facebook, with a much smaller model and much less compute used to train it. Them boys is gangsta! (see pic below.)

Startup Databricks releases world’s most powerful open-source AI

Zoom announces new Workplace AI assistance

Zoom zoomed to prominence during the work-from-home era of the pandemic, because it was widely available and good enough. Now folks are working hybrid, and corporate tools such as Microsoft Teams are growing in prominence. To keep up, Zoom is trying to add some AI sizzle to its plain-vanilla videoconferencing, with a suite of tools they are calling Workplace. Workplace includes the usual bells and whistles such as meeting transcription and summarization, plus other collaborative tools. Seems like Zoom wants to be Microsoft Copilot. The life-or-death question for Zoom is: does Microsoft Copilot want to be Zoom+?

Zoom announces new Workplace AI assistance

OpenAI gives Sora video app to filmmakers - the results are wild

OpenAI CEO Sam Altman has been chatting up Hollywood A-listers. And not just to try to get dates - he is, by all accounts, recently and happily married. Sam wants OpenAI to fuel the AI revolution in filmmaking, so he has been giving key filmmakers access to the company’s boffo new text-to-video app, Sora (see our February 21 issue’s top story.) The results are startlingly original. See for yourself at the link below.

OpenAI gives Sora text-to-video app to filmmakers - the results are wild

AI robot spots sick tulips in Netherlands fields

AI is already starting to transform agriculture. Few industries have to track so many critical variables in near-real time in order to achieve success. One small but lovely example of this trend is the spread of AI robots in the tulip fields of the Netherlands. The autonomous robotic vehicles use AI image processing to examine each individual tulip for discolored spots that indicate disease. Sick plants are pulled from the field to decrease spread of the disease to other tulips, thus optimizing yield per planted hectare.

AI robot spots sick tulips in Netherlands fields

AI designs antibiotics active against pan-resistant pathogens

Modern medicine has caused the evolution of ever-more resistant bacteria. Unfortunately, the development of new, effective antibiotics has lagged far behind in recent decades. This has given rise to fears of a looming crisis in which we may become relatively powerless against pan-resistant superbugs. The World Health Organization has made a top priority of developing an antibiotic effective against Acinetobacter baumannii, a pan-resistant gram-negative bacteria increasingly found in health care sites. Now a team from Stanford in the US and McMaster University in Canada have developed a generative AI system that designs candidate molecules that are both likely to be effective and easy to synthesize. In a recent paper in Nature, they detail how the system designed 58 candidate antibiotics for use against A. baumanii, and 6 were found to be effective.

In the words of the authors, “This demonstrates the potential of generative artificial intelligence to design structurally novel, synthesizable and effective small-molecule antibiotic candidates from vast chemical spaces, with empirical validation.”

AI designs antibiotics active against pan-resistant pathogens

Nvidia partners with Hippocratic to develop $9/hour AI nurses

Leading AI chipmaker Nvidia has partnered with AI startup Hippocratic to develop an AI chatbot that can offer medical information and triage to patients over video in real time. A test of the system was rated by nurses and doctors as equal to the performance of a human nurse on 4 out of 5 rated criteria, and as decidedly superior on Patient Education/Motivational Interviewing (see chart below.)

Nvidia partners with Hippocratic AI to develop $9/hour AI nurses

AI-identified psychedelic may undergo clinical trials

Biotech startup Mindstate Design Labs has filed an Investigational New Drug application for a clinical trial of its candidate psychedelic drug, 5-MeO-MiPT, nicknamed Moxy. The drug was selected by an AI model that searched through 70,000 “trip reports,” narratives of experiences under the influence of various psychedelic substances, drawn from all over the internet.

Psychedelics are having a mainstream moment, with Lykos Therapeutics on track to have MDMA (the club drug known as “ecstasy” or “molly”) approved for treatment of PTSD. AI is now assisting that transformation.

AI-identified psychedelic antidepressant may undergo clinical trials

UK’s NHS AI spots tiny breast cancers missed by clinicians

An AI system tested by the UK’s National Health Service detected tiny breast cancers in 11 women that were missed by human radiologists. The system, known as Mia, analyzed mammograms of 10,808 women. It successfully noted all of the patients found to have tumors by physicians, but in addition found 11 cases missed by the doctors. Mia was able to detect tumors as small as 6 mm, where treatment leads to a greater than 90% 5-year survival rate. Currently in the UK, all mammograms are read independently by 2 radiologists. If Mia continues to be successful, the NHS plans to substitute the AI for one of the radiologists, effectively doubling the number of mammograms that can be read.

UK’s NHS AI spots tiny breast cancers missed by clinicians