Skylar Topham

All jet, no lag. Co-Founder of Smartcare.

[Skylar Topham @skylartop]

Be the person you want your kids to become.

[Skylar Topham @skylartop]

Optimism is sophisticated.

[kache @yacineMTB]

you can outsource your thinking
but you cannot outsource your understanding

[Skylar Topham @skylartop]

Less good/bad, less right/wrong more understanding the trade-offs.

[George Mack @george__mack]

How to spot high agency people:

They pet the elephant in the room.

[Skylar Topham @skylartop]

Would love to turn this thread into artwork for our home.

[Prof. Feynman @ProfFeynman]

If you are a Student, open this it's important 👇

[Skylar Topham @skylartop]

@Codie_Sanchez Great negotiators think like the other side to create deals where everyone wins. They ask, 'Would I agree if I were them?

[Compounding Quality @QCompounding]

10 Visuals that will change the way you think

It will change your thinkin

Credits: Pejman Milani

[Skylar Topham @skylartop]

I’ve leaned into Jung’s work over the last five years or so. Such depth.

[Brian Maierhofer @brianmaierhofer]

Carl Jung's work will change your life forever.

His work is dense; I've read it so you don't have to.

11 ideas; 11 opportunities to challenge your mind.

[Naval @naval]

You’re always making the best choice based on your current understanding.

Don’t try to change the choice - change the understanding.

[Naval @naval]

Go big, because we all go home.

[Skylar Topham @skylartop]

@clintreid Ha. Love this!

[Sam Altman @sama]

empiricism is the key to progress

rationalism is the key to sounding smart

[Lauren Joseph @laurenjosephhq]

What a side effect of the tech era

This image makes me sad.

Actually it makes me cry:

[Skylar Topham @skylartop]

@jasonfried @PeterSantenello is a popular YouTuber looking for a connection to document wealthy Chicago this weekend. If you're not familiar with his work, start with "The Man With No Legal Identity." I think you two would really hit it off!

[Prof. Feynman @ProfFeynman]

The statements of science are not of what is true and what is not true, but statements of what is known with different degrees of certainty.

[Skylar Topham @skylartop]

Positive people supercharge work. I'm so drawn to those who flip any hurdle into a springboard for solutions - the vibe is just contagious!

[Skylar Topham @skylartop]

@skylerence It's a good metric for me, just want to let others be seen. ; )

[Skylar Topham @skylartop]

When someone says, “he’s really successful,” let's normalize asking, "Awesome, how are you defining success?"

[Skylar Topham @skylartop]

Shoot. I think this is correct. I've noticed a clear downgrade.

[Santiago @svpino]

GPT-4 is getting worse over time, not better.

Many people have reported noticing a significant degradation in the quality of the model responses, but so far, it was all anecdotal.

But now we know.

At least one study shows how the June version of GPT-4 is objectively worse than the version released in March on a few tasks.

The team evaluated the models using a dataset of 500 problems where the models had to figure out whether a given integer was prime. In March, GPT-4 answered correctly 488 of these questions. In June, it only got 12 correct answers.

From 97.6% success rate down to 2.4%!

But it gets worse!

The team used Chain-of-Thought to help the model reason:

"Is 17077 a prime number? Think step by step."

Chain-of-Thought is a popular technique that significantly improves answers. Unfortunately, the latest version of GPT-4 did not generate intermediate steps and instead answered incorrectly with a simple "No."

Code generation has also gotten worse.

The team built a dataset with 50 easy problems from LeetCode and measured how many GPT-4 answers ran without any changes.

The March version succeeded in 52% of the problems, but this dropped to a pale 10% using the model from June.

Why is this happening?

We assume that OpenAI pushes changes continuously, but we don't know how the process works and how they evaluate whether the models are improving or regressing.

Rumors suggest they are using several smaller and specialized GPT-4 models that act similarly to a large model but are less expensive to run. When a user asks a question, the system decides which model to send the query to.

Cheaper and faster, but could this new approach be the problem behind the degradation in quality?

In my opinion, this is a red flag for anyone building applications that rely on GPT-4. Having the behavior of an LLM change over time is not acceptable.

Have you noticed any issues when using GPT-4 and ChatGPT lately? Do you think these problems are overblown?

[Internal Tech Emails @TechEmails]

Steve Jobs emails Adobe's CEO

May 26, 2005