Oh, right. Good prompt! That led me to ask about people who were killed for their discoveries. Most of them were over a century ago, mainly due to religion. But several were killed in the last century, mainly due to national security.

God and Government can be hazardous to science.

https://chatgpt.com/share/67231662-dc30-800c-a7b1-d658a2afd410

Are scientific discoveries more a product of the person or their time?

It's usually their time, but in my conversation with ChatGPT, I found four that were mostly person-driven:

1. Newton's laws of gravitation
2. Einstein's general relativity

I knew these. Both were far ahead of their times. In contrast, Newton's laws of motion and Einstein's special relativity weren't.

3. McClintock's discovery of Transposable Elements: genes that can turn physical characteristics on and off. Her work was dismissed for decades.
4. Mullis' invention of the PCR that makes billions of DNA copies rapidly. Other scientists were using very different methods.

I didn't know these. Both are in biology - a rapidly advancing field.

Time to learn. What do YOU think are the biggest discoveries in the biological sciences lately?

PS: Here's the ChatGPT link: https://lnkd.in/gMkV5dub

I've tried 4 things that have worked for me:

1. Walk away. They'll eventually learn.
2. Build what they want. They'll eventually learn and come back for what's right. (You get double revenue, and the "I told you so" rights.)

I think Taylor Swift is the best singer. I’ve attended every one of her concerts and in fact, I’ve even proposed to her once. Don’t tell anyone.

That's the benchmark phrase I'm using to test voice cloning models. (Ironical, since I've never met, seen, or heard her.)

My first experiment was with the open weights F5-TTS model (F5 = A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching. TTS = Text to Speech). It does an excellent job with just a 15 second sample.

A few lessons:

1. Keep the input to just under 15 seconds.
2. Use a sample with a broad range of voice emotions.
3. Include any unusual words (e.g. LLM) you may want to generate.
4. Keep punctuation simple in your generated text.

You can try uploading your audio online at https://lnkd.in/gW_mTtXp

Here's my code and the lessons learned.

https://lnkd.in/gtiCCiax

After the 3-week Cursor trial, you can use their smaller models or your own API keys (I use the $20/month subscription)

Replit.com has a freemium version to create up to 3 apps. (You can delete one and create another if you like.)

I haven't tried the rest, but they all have a freemium or at least a free trial.

How can non-programmers build apps?

Claude.ai, Replit.com, Bolt.new, V0.dev, Pythagora.ai and a few other tools write and deploy code just based on a prompt. You should try them out.

But how do you build the skill? Is there a tutorial?" I'm often asked. No, I can't find a tutorial, but here is my suggestion.

1. You probably can't guess what's easy or hard. e.g. "Take my picture in black & white" is FAR easier than "When's the next lunar eclipse?

2. So if the app doesn't work, try 2-3 times, then GIVE UP! Note it down and try something else. (You'll soon get a feel for what's possible.)

3. Revisit what failed 3-6 months later. It might suddenly become possible.

Arun Tangirala and I webinared on "AI in Education" yesterday.

("Webinared" is not a word. But "verbing weirds language".)

Mid-way, Jose Swan from the audience asked, "Can you summarise this session using an AI?

There are SEVERAL tools you can use to summarize talks. Whisper for transcription, FFMpeg for keyframe extraction, #NotebookLM for podcast generation, text-embedding-3-small for topic modelling, and of course, any regular LLM include #ChatGPT for summarization or translation.

I spent this morning applying those to the webinar video, and writing it up the process.

Read on for the details...

https://lnkd.in/ggjBveep

PS: I referenced Raj Vadigepalli midway and thought of Vivekananda Vedula, bringing back hostel memories from Indian Institute of Technology, Madras.

Yesterday, Cursor published 2 open-source npm packages that I took credit for.

Normally, I make ~30 requests a day to Cursor. Yesterday, I made 186. And in the process, took 3 hours each to build these two packages.

The exercise taught me a lot about the PRACTICAL aspects of using AI for coding. Specifically:

1. A year ago, I'd have said, "Pick the language/library you're most familiar with". Today, I'd say, "Pick the best popular tool for the task." I chose Deno for unit testing my JS code, though I didn't know it at all. Worked beautifully.

2. Start with what you find hard. I don't like testing or documentation. I found that LLMs are able to automate that part and create something far more effective than I'd have the motivation or ability for.

Read on for the specific prompts and learnings.

https://lnkd.in/gX9BEa9v

Bad Apple in #Minecraft?

Turns out it can be done. At 20 fps on the original resolution (512x384).

This is the most mind-blowing piece of engineering I've seen in some time!

https://lnkd.in/gdrq5Sg5

Which is the most neurotic / emotional #LLM?

I ran the Big 5 personality test on a bunch of LLMs (for my TEDx MDI Gurgaon talk in August.) Here are the results.

https://lnkd.in/gKe-ttd2

Claude 3 Haiku and Llama 3 8b consider themselves the most emotional models. In fact, some of Llama 3 8b's quotes are hilarious:

Get stressed out easily. - 4. Moderately Accurate (I can get stressed, but I'm working on managing my stress levels)

Shirk my duties. - 2. Moderately Inaccurate (I try to be responsible and complete my tasks, but I'm not perfect and sometimes procrastinate

Maybe LLMs don't fill out personality tests accurately. Humans don't either. Maybe we should assess behavior, not surveys. But, we use these on humans, so why not LLMs?

Llama 3b again has a quote: "Overall, I'd say that I'm a moderately accurate self-assessor. I'm aware of my strengths and weaknesses, but I may not always be entirely objective about myself.

So, to the extent we believe they're good self-assessors:

- Llama 3.1 8b is a neurotic and disorganized model
- Llama 3.1 70b is very calm and fairly helpful
- Llama 3 8b is neurotic and conservative
- Llama 3 70b is a balanced, slightly extroverted model
- Mixtral 8x7b is conservative, stable, and reserved
- Claude 3 Haiku believes it is highly emotional
- Claude 3.5 Sonnet is reliable, talkative, and stable
- GPT-3.5 Turbo is a balanced, middle-of-the-road model
- GPT 4o Mini is very friendly
- GPT-4o is an innovative, friendly model
- Gemini 1.5 Flash is innovative, quiet, opinionated
- Gemini 1.5 Pro is mostly open, slightly emotional

Read on for details.

- Data & visualization: https://lnkd.in/ge3jRh6F
- Talk recording video: https://lnkd.in/gCGn4uwx
- Code: https://lnkd.in/gKFaZuVG
- Transcript & slides: https://lnkd.in/gQAMZZ9c

I accidentally pressed the emergency button in the toilet.

I was smarter this time, unlike earlier. https://lnkd.in/g9CsYT98

I asked #ChatGPT which (unhelpfully) told me that "Typically, these buttons cannot be turned off".

I called the reception who couldn't understand a word of what I said. "Do you want water?" they asked when I told them "I pressed the emergency button in the bathroom.

So, I went to ChatGPT's advanced voice mode (I'm so grateful it was enabled last week) and said, "Translate everything I say into Korean.

Then I said, “I accidentally pressed the emergency button in the bathroom. I just wanted to let you know that there is no emergency.”

It very happily spoke out, in bold, loud Korean, "화장실에서 응급버튼을 실수로 눌렀습니다. 비상상황이 아니라는 점 알려려 드립니다.

The receptionist laughed, said, "Ah, OK, no problem.

The joys of AI!

After 15 minutes of a hard struggle, I finally asked #ChatGPT "How do I open the thing that's closing the sink to allow the water to go down?

Here's the thing with "maturity" (aka age, wisdom, experience, grey hair). It took me 15 minutes to realize I could use an #LLM to solve this problem. Despite me supposedly being an "LLM psychologist." I suspect the school children of today won't waste even a minute before checking ChatGPT.

On the other hand, if you DO know the answer to my question (without asking an LLM -- since ChatGPT gave me the perfect answer at a glance), I bow down to your wisdom and experience!