Rich Whitcomb

Nytimes is a cooking and games company that helps support a small journalism arm on the side sadly.

Is it limited in pro or normal subscription? I have it going on something now but didn’t realize it had a low limit.

I enjoy it so far but it feels closer to Diablo 4 than Baldur’s Gate.

Curious if you also played the new dragon age? Looks somewhat similar genre at a high level.

I haven’t logged into Twitter in awhile. Looks like the algorithm has been tuned into right wing TikTok with Temu ads?

I found a nice boilerplate project that used it and was able to push a new library to pypi in like 20 minutes. Amazing how it easily drops in place.

This post sent me down a small rabbit hole reading about that cyclic functional group on the bottom. Looks like it exists in nature?

Does Accenture even know what Accenture does?

I haven’t used Devin directly but feels similar to my experience with other tools.

Performance similar to an intern who did very well on their college classes but with zero experience how coding works in the real world.

That would be a fun recsys challenge. Look for something like a percentage of your feed to be close to the same embedding vector and then hide posts within a distance to that vector into the future for N days.

Social media needs an unsubscribe button like email: “I’ve read every possible take on this trending story. Please hide all future posts about it.”

We’ve hit the point where a lot of the tech from sci-fi movies is possible but makes almost no sense to build in the real world

I always do this: dive into a new framework or language, build something impossible to maintain, then start over with a new one every two years.

High pressure to send quality content to that one person.

To discuss applying for and studying in LLM programs This subreddit is a place for people who are considering studying for a LLM, are in a LLM program, or have finished a LLM. The focus is on LLMs in the US and UK, but discussions on all LLMs are welcom...

The law LLM subreddit being taken over by large language model is also a funny result of this mix up.

www.reddit.com/r/LLM

The worry is that the app can exploit a zero day vulnerability of android or iOS to escape the sandbox and track location and other usage of the phone at the direction of a Chinese govt agency.

Supreme Court wasn’t convinced by the propaganda argument either.

Even a well funded company, with thousands of researchers, can’t stay that far ahead the other billions of people on earth.

How about most social media apps stop with all of this chatbot integrations and instead help me with simple spelling mistakes before I post…

Sad that a company with the name of “OpenAI” might have fundamentally killed off the “research in public” philosophy the tech industry has had since the early 2000s.

The closed nature of LLM research makes me feel this way. When reading a paper in the field you constantly have a thought that the big companies have advanced beyond this insight.

One fear I have with AI coding is that we will become complacent with the quality of the libraries and abstractions in our tools and coding languages.

Most code generation is helping with bad library and PL design. Looking at you matplotlib in particular.

They also forgot you have to hire more people so there is actually someone to hand you the product.

How goes gmail have all of these useless fancy LLM features but nothing that screams at you if you accidentally misspell someone's name in the body of an email.

In the case time to lean heavily into the neon biking gear. Bonus points if you wear it hiking.

Lot of the crypto and AI wave has been folks rediscovering basic abstractions and feeling like they invented something new.

Maybe this is just every tech cycle and I have just been around ling enough now to have been through a few.

I still don't understand agents. Is any code that calls a LLM API and makes a decision with the output an agent? Have I been coding agents for years?!

Working with LLMs after years of coding is… frustrating. With traditional coding, the computer does exactly what you tell it—a blessing and a curse.

With LLMs? You’re left guessing what they’ll do, even while debugging.

Anyone wants a very unsexy idea for an "AI" company look into appliance part replacement...

Something so very Las Vegas that the new A's stadium in the city will have a view overlooking a fake NYC skyline.

I agree but also the situation was not great before the Gen AI summeries. Can't tell yet if this is a worse situation. Depends if people believe the summary more than the bad link.