Weekend Briefing No. 1
Welcome to my first weekend briefing newsletter. It's a curated list of links and information I believe will entertain, inform, and be of use to you.
Interesting Data Points
- Why is it bad to go to a casino? Understanding the law of large numbers
- Cost of living, unemployment, and climate change are top worries of Gen Z and Millennials
- Are SVMs better at text classification than pre-trained language models?
AI Chatbots
A few newsletters ago I posted a link to an article about Wendy's wanting to replace their drive-thru order takers with a GPT chatbot. They're not the only ones with this idea. The National Eating Disorder Association (NEDA) has decided to replace all their human helpline staff with an AI chatbot.
It's believed that this chatbot was deployed as a union-busting tactic because NEDA workers unionized in May.
"Staff were informed of the change, and of their firing, just four days after they successfully unionized, according to a blog post written by helpline associate and union member Abbie Harper earlier this month."
Was NEDA successful in deploying a chatbot to replace their human workers? Not at all, it appears that their chatbot "Tessa" failed miserably.
"As of Tuesday, Tessa was taken down by the organization following a viral social media post displaying how the chatbot encouraged unhealthy eating habits rather than helping someone with an eating disorder. "
Chloe Xiang
I'm not sure why so many companies are planning to replace workers with chatbots. Do they think it's a good idea? It might be a good idea for low-level repetitive work like order taking for Wendy's example but for a helpline? Bad mistake NEDA!
Nvidia skyrockets
There's a saying about making money in a gold rush is not in the gold itself but in selling the shovels. LLMs are all the rage right now and Nvidia (NVDA) is selling the shovels.
NVDA shot up almost $100 per share on Thursday, May 25, 2023, as it rides the AI wave.
NVIDIA is now valued at 1 trillion dollars and to keep the party going, it introduced a beast of supercomputer - the DGX GH200. It has 144 terabytes of shared memory and 256 Grace Hopper Superchips acting as a single GPU. This is one helluva beefy machine.
I had to read up on the Grace Hopper Superchips and understand them to be a powerful CPU and GPU high-performance computing chip combination. I wonder if it can run Fortran?
It's only fair to follow up the above NVIDIA announcements with an understanding of the differences between a CPU and a GPU, and which is faster at calculating prime numbers.
Dave's Garage does a wonderful job of comparing and contrasting these chips and helps us understand where each chip type shines.
AI stock picker
Having AI pick stocks isn't anything new but now it's GPT powered. JPMorgan released IndexGPT last month.
"The company applied to trademark a product called IndexGPT this month, according to a filing from the New York-based bank."
This is merely another evolution in the "robo-advisor" field with GPT providing customers more information and options to trade. I think this type of GPT application can be very beneficial for both customers and wealth advisors.
This might qualify for low-level repetitive work where automation with a bit of "smartness" built in might work well. I just wouldn't put it in charge of a hedge fund just yet.
AI will kill us all
Are we close to the Terminator world? Some AI researchers are really concerned that GPT will kill us all in some way. My worry has always been how it can be used for disinformation and manipulation.
"“Mitigating the risk of extinction from A.I. should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war,” reads a one-sentence statement released by the Center for AI Safety, a nonprofit organization. The open letter was signed by more than 350 executives, researchers, and engineers working in A.I."
A recent fake image shared by a verified Twitter account (thanks Elon) caused quite a bit of confusion until it was caught and the reports rectified.
“This image shows typical signs of being AI-synthesized: there are structural mistakes on the building and fence that you would not see if, for example, someone added smoke to an existing photo,” Hany Farid, a professor at the University of California, Berkeley, and digital forensic expert told CNN.
Here's my take on the future job market; become a digital forensic expert! We'll need a lot more of them in a GPT world.
End Notes
If there's one sentence I could write that summarizes this week's newsletter it's this: GPT isn't going away and will continue to grow as more powerful GPUs hit the market. Love it or hate it, GPT will transform our world in the future. The caution is, how much and how far will we let it go?
Member discussion