AI Top-of-Mind for Nov 20
Top-of-mind, there have probably been a hundred articles over the past weekend concerning OpenAI, Sam Altman, and Microsoft. CNN had a good analysis of what led to his ouster, the structure of OpenAI, and the potential next steps. From the article:
· OpenAI’s overseers worried that the company was making the technological equivalent of a nuclear bomb, and its caretaker, Sam Altman, was moving so fast that he risked a global catastrophe.
· That probably led Altman to push the for-profit company to innovate faster and go to market with products. In the great “move fast and break things” tradition of Silicon Valley, those products don’t always work so well at first.
· That’s fine, perhaps, when it’s a dating app or a social media platform. It’s something entirely different when it’s a technology so good at mimicking human speech and behavior that it can fool people into believing its fake conversations and images are real.
And as of today, he joins Microsoft, along with Greg Brockman, with Emmett Shear appointed as OpenAI’s interim CEO.
A great article from the HBR on how marketing and sales can properly embrace Generative AI. From the article:
A recent paper in Science predicts that generative AI will also have a transformative impact on the sorts of complex writing tasks that are the bread and butter of many marketing jobs — for example, sales support, marketing materials, websites, content marketing, and social-media posts. When the authors of the study gave study participants access to ChatGPT, the amount of time they needed to complete a set of writing tasks decreased by around 40%, and the quality improved by about 20%.
Some conclusions from the authors:
· Invest in training.
· Embrace an intentional test-and-learn approach.
· Establish policies to safeguard customer and proprietary company data.
· Ensure quality control.
· Enhance creativity, don’t replace it.
Also on the marketing front, an AI-powered ‘Insights Machine’ that enables marketers to interact with gathered data and even create virtual focus groups. The Article is behind a paywall, but here is a link to the agency’s AI site.
Another follow-up further explaining OpenAI’s announcements from a week back, an older tutorial on developing effective GPT prompts, and ‘The Verge’ describes new AI-powered text and video editing coming to Instagram and Facebook.
Finally, more on spinning up LLMs on your PC or laptop by Benjamin Marie in ‘Towards Data Science.’ For those interested:
In this article, I explain how you can run Falcon-180B on consumer hardware. We will see that it can be reasonably affordable to run a 180 billion parameter model on a modern computer. I also discuss several techniques that help reduce the hardware requirements.
