Humanity's Destiny

Plus: Llama 3 dominates GPT-4, Gen Z chooses blue collar jobs, Sam Altman on AGI and much more...

Hey everyone,

A special welcome this week to our new subscribers—great to have you here!

This is your Sunday Space, where I serve up the best ideas, tools and resources I’ve found each week as we explore the technology shaping the future.

If you find something thought-provoking, forward it to a friend.

Humanity’s Destiny

Created with Midjourney

In Frank Herbert’s Dune universe, humans waged a violent war against the machines which oppressed them.

In Isaac Asimov’s Foundation universe, another famous sci-fi series, AI secretly manipulates society, leading to their ultimate dissociation from humans.

In both series, humans and machines do not have a happy ending.

Of course, these are just speculative, fictional stories imagined (brilliantly) by their authors.

But as our society hurtles toward a future of AGI—one where the “thinking machines” of Dune become a reality—I can’t help wondering whether sci-fi has been prescient, warning us of the consequences of such a creation.

I have no doubt the entrepreneurs and inventors forging this path are well aware of the risk. After all, their ventures were inspired by sci-fi.

As an optimist, I believe technology will lead to humanity's prosperity and preservation for millennia.

That doesn’t mean we can sit back and relax with blind hope.

As the builders shaping the future, whether you’re a prompt engineer or the engineer creating AGI, we must do what we can to ensure it benefits humanity instead of destroying it.

Llama 3’s Dominance

I recently wrote about how Llama 3 claimed its title as the undisputed champ of open-source models.

After digging into it more, it’s incredible to see how fast and cheap its 70B parameter version is compared to the best (closed-source) models, particularly:

  • GPT (OpenAI)

  • Claude (Anthropic)

  • Gemini (Google)

While Llama 3 (the blue dot) lags GPT-4 Turbo and Claude 3 Opus in quality, it smashes them in throughput (i.e. speed in generating answers) and price—the metrics that, I believe, will ultimately matter to the end user.

Notice how no model has reached the promised land (the green quadrant) yet…

Quality (i.e. smartness), speed and cost are the holy trinity of AI models.

No matter how smart a model is, users won’t care about it if it’s expensive and slow to generate answers.

AI Word of the Day


The variables inside a large language model (LLM) like GPT-4 (the model powering ChatGPT). Generally, the more parameters a model contains, the smarter and more sophisticated its output.

I’d love to hear if this AI Research segment interests you and what you want me to cover. Let me know in the poll at the bottom of this email or on X/Twitter.

From around the web

Out of fear of AI-fuelled job displacement and wanting higher earnings without the burden of college debt.

This, and Gen Z’s revival of the long-forgotten landline, fits into the broader trend of their resistance to being too connected and reliant on their mobile phones. Personally, I’m a big fan of this.

I find it a fascinating paradox, with the inevitable emergence of AI and robotics set to drive humans toward more knowledge work.

But, it seems the trend towards more blue-collar careers is likely to accelerate as Gen Z continue to shape the workforce of the future.

Luca’s take: I think Gen Z has a massive opportunity. Imagine supercharging a boring plumbing business with no code and AI tools. I have no doubt this generation will continue their trailblazing well into the future.

The CEO of OpenAI, never afraid of public appearances, shared some insight on his company’s current models and the path to AGI:

  • He called GPT-4 "mildly embarrassing at best", saying it will be the "worst model" we will ever use as each new version gets smarter.

  • He “doesn’t care” whether the company burns $500M or $50B annually. As long as it builds AGI, it’ll be worth it.

  • Altman also spoke about the importance of global access to compute, clearly communicating their mission to make ChatGPT free for as many people as possible.

Luca’s take: as someone who marvels every time I use GPT-4, I still find it hard to imagine that we’ll look back at it as the worst model ever. Things are going to get crazy when sama drops GPT-5. Maybe this is his way of building some initial hype, not that it’s needed.

Their suite of AI models fine-tuned for healthcare use cases.

Source: Google

Although benchmarking for AI models is becoming increasingly scrutinised, Med-Gemini beats GPT-4 in 10 out of 14 different medical tasks, even incorporating web search for more reliable and up-to-date outputs.

Luca’s take: I’m a sucker for vertical AI solutions that will drive enormous productivity gains in every industry. I can’t wait to hear how many more patients Med-Gemini helps doctors treat worldwide.

On X

  • “Cash-flow is an exit very year…”

On YouTube

The No-Code x AI Bootcamp

If you're a regular here, you’ll know I’ve been through the Bootcamp myself. I'm not big on courses, and I recommend even fewer.

That speaks volumes about this one. It was a game-changer.

I’ve always been 'almost there.’ Full of ideas, yet stuck at the starting line, watching others on Twitter launch their products and succeed.

Then, I joined the Bootcamp. It completely changed my perspective.

I discovered the true power of building with No-Code and launched my MVP in a weekend.

I’m no longer hampered by analysis paralysis and have a roadmap of ideas and products I plan to build.

If you’re feeling stuck like I was, this Bootcamp is the way to break free. The May 2024 cohort kicks off tomorrow (May 6th), so this is your last chance.

Quote I’m Pondering

"The dangers that face the world can, every one of them, be traced back to science. The salvations that may save the world will, every one of them, be traced back to science.”

—Isaac Asimov, Author of the Foundation science-fiction series

What did you think of today's edition?

How can I improve? What topics interest you the most?

Login or Subscribe to participate in polls.

Was this email forwarded to you? If you liked it, sign up here.

If you loved it, forward it along and share the love.

Thanks for reading,

— Luca

P.S. A list of things AI can’t do… yet.

*These are affiliate links—we may earn a commission if you purchase through them.

Join the conversation

or to participate.