AI Top-of-Mind for 7.2.24 — ‘Little Tech’
Today: Big tech and open-source, Character.AI, an update on AI data center energy consumption, Google’s new Mesop, and are we biased against AI-generated music?
Top of mind is the power of ‘big tech,’ with ‘The Information’ diving into the importance of open-source and ‘little tech.’ From the article:
· The U.S. — the world’s innovation hub and leader in cutting-edge artificial intelligence — risks ceding its leadership through policy missteps driven by Big Tech’s outsize influence. In Washington and across the country, Big Tech wants to shape regulations to extend its dominance. That includes efforts to restrict the development of open-source technology, a key driver of innovation that will help determine the global leader in the development, deployment and application of AI.
· Seizing this moment to fully realize AI’s benefits will require a regulatory environment that advances a few basic principles — similar to the early days of the internet: We need to prioritize the manufacturing of AI technology and software in America, and open-source AI models should be allowed to compete fairly. To ensure that Little Tech has a fair shot, we need to avoid regulatory capture and market consolidation. We should regulate the applications of AI rather than the underlying technology.
Note that I was previously in a company where we homed in on the same issue, looking at the incumbancy of ‘Big Telco’ and comparing it to Big Oil, Big Pharma, and Big Tech. An earlier chart by ‘CB Insights on market consolidation:’
On the model front, a new Python UI framework from Google. Jim Clyde Monge writing in ‘Generative AI’ looks at ‘Mesop,’ a Python-based UI framework that enables rapid development of web applications, particularly demos and internal tools. He highlights its advantages: flexibility and customization, ease of use, and the ability to handle complex projects. Example of output for a text-to-image web app:
Also, from ‘The Information,’ potential partnerships between Character.AI and both Google and Meta. From the article, problems with the company that pioneered character-driven bots such as TV personalities or historical figures:
· As some of the novelty of personalized chatbots has worn off, competition has intensified. Meta Platforms last week started testing chatbots made by individual creators after it released chatbots modeled after celebrities like Tom Brady last fall. Google has also been developing a product for creating and conversing with custom chatbots, which could launch as soon as this year.
· But many of these startups have been grappling with the high costs of training and running AI models, plus competition from tech giants and large startups such as OpenAI. Few AI entrants have shown they can make enough money to offset the costs.
There has been plenty of coverage as to AI data center energy requirements. Here, Enrique Dans tells us not to panic given proven energy efficiencies due to AI tools as well as the placement of data centers where renewables are available. He links to a good older article that includes metrics on Google search energy consumption. A recent Forbes article notes:
Data centers supporting AI alone may well end up using a total of 1,000 TWh annually by 2026, or roughly the equivalent of the entire annual electricity consumption of Japan — a country with a population of 125 million people.
And from the same article, a recent talk by Bill Gates where he offers the following addressing the above prediction:
· Data centers will drive a rise in global electricity usage of between 2–6%, Gates said. But added: “The question is will AI accelerate a more than 6% reduction? And the answer is: certainly.”
Onto creative, Sophia Omarji in ‘Illumination’ asks the following question: Can our brains tell the difference between AI versus human-composed music, and can AI music evoke the same emotions as that created by humans? The gist of the article is that, if people don’t know a given piece was AI-generated, they’d feel the same connection as something written by a human composer. However:
Whilst AI can mimic existing patterns to create something new, human creativity is unique in the sense it is driven by emotional depth, personal experience and intention. Here, AI replicates the process of creativity, but the authenticity of human artistic expression remains distinct.
Lastly, on PR, an unbiased look by ‘PR Daily’ at how enterprises are adopting AI and the challenges they sometimes face.
· Indeed, the Washington Post pointed to a litany of delayed launches and broken promises on AI technology, ranging from Amazon still not delivering on its AI-powered Alexa revamp nine months later to Microsoft pumping the brakes on AI integrations into the Windows operating system to Google backpedaling on some of its AI-driven search results.
· The column cautions that, “it’s important to recognize that your imagination may outpace the capabilities of AI.”
· And finally, despite all its trials and questions, AI is changing the workplace today. One early winner? Consultants. Boston Consulting Group told the New York Times that one-fifth of all its work is now related to AI — up from zero just two years ago.
