AI Top-of-Mind for 3.25.24 — Market Landscape
Top-of-mind is really the ‘Three Body Problem,’ but not for here.
Top-of-mind on the AI front is a great update from A16z on Gen AI app momentum. Pretty comprehensive analysis that spans the globe and breaks out desktop vs mobile. Not many surprises, but bunches of apps I’ve never heard of, including all the companion apps.
On the practical front, a follow-up to an older article on the use of AI for antibiotic development. Dr. Ashish Bamania in ‘Level Up Coding’ offers a thoughtful analysis on how it is becoming increasingly difficult to develop bacteria-resistant drugs and where AI can help. From the posting:
• For a long time, antibiotics have been discovered by studying and screening soil-dwelling microbes to look for chemicals that they produce against bacteria that might kill them….
• However, this technique is becoming less effective, as researchers frequently find antibiotics that are structurally similar to those already known, making it challenging to discover new and unique compounds….
• This exact problem was solved using AI, specifically by using a Directed-Message Passing Graph Neural Network (D-MPGNN).
At the end of last week, I wrote about Microsoft’s ‘absorption’ of Inflection. The company’s goal in introducing the ‘Pi’ model as to offer something more conversational. Unfortunately, as Ignacio de Gregorio details, it never got the necessary traction.
And now three items on the creative front. If you’re not a great artist, like me, a great way to turn simple sketches into real artwork with Midjourney. More from Diana Dovgopol in ‘Artificial Corner.’
And we’ve also spoken also about ‘Midjourney’ Character Reference. Now we have Style Reference, and results as described by Jim Clyde Monge in ‘Generative AI’ are below.
And finally, new prompt techniques for ‘Leonardo AI.’ Corbiz in ‘Artificial Intelligence in Plain English’ goes into noise generation, prompts, and strange artifacts.
Turning to the enterprise, an update on digital twins. ‘EE Times’ details how the NIST is adopting twins for its new semiconductor manufacturing institute, also extending the use for actual facilities planning and leveraging AI.
Also on the technology front, but not as positive is a report that China could be creating backdoors in PCBs shipped to the US for vital sectors like utilities. A chilling quote from the article:
• “You can’t put together a power grid without relying on Chinese circuit boards or non-domestic circuit boards,” Al Shaffer, former deputy undersecretary of defense, told EE Times. “If you rely on non-domestic circuit boards, and there’s a remotely triggerable kill switch, in essence, the potential adversary could control your power grid.”
• “I look at the recently passed CHIPS Act, and I think they made a real fundamental mistake. The government is putting $50 billion into the production of semiconductors. Semiconductors don’t make anything work. You need to have them mounted into a system. We’re spending $50 billion to increase domestic capacity, and then we’re going to go mount these things on primarily Chinese- or Taiwanese-, Korean-made PCBs. A lot of those will be Chinese, because the Chinese have the largest share of state-of-the-art [boards]. We also know China uses microelectronics systems for espionage.”
And then back to the enterprise, ‘CIO’ reports on Accenture’s view that a ‘people-first’ approach to AI is critical. From the article:
GenAI is capable of reinventing the very nature of work, reshaping how businesses deliver value and better experiences for employees and customers. But there’s a trust gap between potential and reality. Accenture’s recent research on the nature of work in the age of gen AI says 95% of workers surveyed say they see value in working with gen AI, and 82% say they already have some understanding of the technology. However, their biggest concern is trusting their employer: 58% say gen AI is increasing their job insecurity and 57% need clarity on what this technology means for their careers.
