Happy Saturday! Here are this weekend's curated links for your reading pleasure.
Interesting data points
- The new Falcon-40B LLM model was released as open source; here are the latest leaderboard rankings of LLMs
- Not good news, adult depression rates are rising
- PFAS manufacturers knew the danger of these toxic chemicals for decades
Is the Earth tipping?
A study published in Nature this past May, 2023 looked at the safe limits for earth system resilience and human well-being. The study found that out of eight categories, we've already at or surpassed the limits for seven of them.
The study by the international scientist group Earth Commission looked at eight scientifically established safe limits, including climate, air pollution, phosphorus and nitrogen water contamination from fertilizer overuse, groundwater supplies, fresh surface water, the unbuilt natural environment, and the overall natural and human-built environment.
The team of forty scientists found that only air pollution was the only planetary limit that wasn’t quite at the global danger point but was instead dangerous at local and regional levels.
I often write about climate-related issues and concerns in my Weathered publication and I believe there's no greater threat to our survival right now than climate change and the degradation of the environment.
In case you want to see where the Earth System Boundary (ESB) transgressions are located:
Object detection at railroad crossings
Before I joined the machine learning / AI revolution in 2014, I worked for a railroad consultancy in New Jersey. Our clients were all the usual public and private rail transportation companies and I designed parking lots, ADA access, and stormwater conveyance systems.
One of my integration points was at railroad crossings and I worked with our "grade crossing" to design and install pedestrian and vehicular safety features. Still, every year there were major accidents involving trains, vehicles, and pedestrians.
An old colleague of mine posted a note the other day on LinkedIn about a company that's incorporated deep learning/object detection into their grade crossing poles and arms. That lead me down the rabbit hole of searching for a research paper on this particular topic.
The PubMed article above is quite a fascinating read, especially how the accuracy of the model for this use case jumped when they applied transfer learning. I believe that deep learning vision applications will have a positive effect on the civil/transportation business in the future as long as the use cases are tightly defined and outcomes measure.
Why AI will save the World?
March Andressen published a positive Substack on how AI will save the world. He makes a lot of great points. I agree, especially right now with the amount of negativity and fear around GPT. I tend to fall into the "we need to put guide rails around it" when it comes to emerging AI technologies before introducing them willy-nilly.
Marc goes on to write a lengthy essay with five AI risks:
- Will AI Kill Us All?
- Will AI Ruin Our Society?
- Will AI Take All Our Jobs?
- Will AI Lead To Crippling Inequality?
- Will AI Lead To Bad People Doing Bad Things?
I think Marc highlights very important questions and I'm a big believer that AI and the associated machine-learning models are a big boon for humanity, if and only if, we use to for the betterment of all humans.
No, that’s not going to happen – and in fact AI, if allowed to develop and proliferate throughout the economy, may cause the most dramatic and sustained economic boom of all time, with correspondingly record job and wage growth – the exact opposite of the fear.
While I agree with his take on AI replacing workers is spot on, I fully believe we need to have government, societal, and cultural safety nets in place for the replaced workers. While it's easy to say that they can be retrained for a new workforce, it takes time and money.
The intersection of technology, society, and culture can be volatile at times. It's only human nature to fear some aspects of AI, especially if your job or industry could be made redundant. The moral of this story is to always be aware of emerging trends and new data points so can make adjustments to your life.