Collaborative Intelligence Week in Review - 19Jul2024
Best business article(s) I read this week…
The author of this piece completed “a quest to go deeper to understand why everybody seems to disagree on exactly what AI is, and why you’re right to care about it.” It provides a different perspective than most articles. I enjoyed reading it.
Harnessing the Value of Generative AI
This is a survey of generative AI’s use across industries from Capgemini’s Research Institute. The report identifies the following four themes, then concludes with a section titled “How organizations can kick-start their generative AI journeys“:
Organizations view generative AI not as a disruptor but as an accelerator
Organizations see more gain than pain in generative AI
Generative AI packs the most punch for IT, sales, and marketing
High-tech leads in implementing generative AI
2024 Technology Trends Outlook
This is McKinsey & Company's latest annual survey. It examines 15 trends, including those related to AI. As always, it’s very well done.
Best technical article(s) I read this week…
SpreadsheetLLM: Encoding Spreadsheets for Large Language Models
A team of Microsoft researchers has released a paper on a new method for encoding spreadsheets for consumption by LLMs. The methods address many problems, such as diverse layouts, inconsistent number formats, etc., that make it difficult for LLMs to discern the content. The researchers claim that using SptreadsheetLLM significantly improves performance on many spreadsheet tasks.
Other items I found valuable/interesting…
Improving the British Open Golf Championship with AI and Digital Twins
If you’re a golf fan, you know this week is The Open Championship. This article describes how NTT Data used a combination of AI, digital twins, and 5G to improve the fan experience.
Transforming Legacy Buildings into Digital Assets
This article describes how historical landmarks and other legacy buildings are preserved with the help of digital twins created via non-invasive data-gathering techniques.
The UK politician accused of being AI is a real person
This is a sign of the times. A UK politician contracted pneumonia, so he missed several campaign events. Rumors spread that he was an AI bot, not a real person. He managed to set the record straight. I included this because it exposes a problem: as AI becomes more powerful, it will become more difficult for humans to trust what they see and hear.
Test-Time Training (TTT) models might be the next frontier in generative AI
A new AI architecture hopes to replace the Transformer architecture. TTT was developed by Stanford, UC San Diego, UC Berkeley and Meta researchers. The team claims that TTT models can process more data than transformers and do so with much less computing power.
Coolest thing I saw…
Soft robots fusing to complete a task
The Faboratory at Yale University has released a research paper and accompanying video demonstrating soft robot amputation and fusion. I don’t know if it’s “cool,” but it’s definitely different and a sign of things to come.
A company that caught my eye…
The company manufactures AI chips using an a “compute in memory” architecture. They recently hired Dr. Taner Ozcelik, founder and former VP & GM of NVIDIA’s automotive division, as its new CEO.