Youtube hearted comments of Martinit0 (@Martinit0).

  1. Hi David, Great post! The official announcement that fusion ignition has been achieved at NIF came just as your video went online. So good nose ;-) The funny thing about laser-based fusion is that the US has a big gap in commercial laser capability: There is NIF, which is a unique government entity and then there is nothing up until relatively 'pedestrian' commercial lasers by companies like Coherent (COHR) and MKS Instruments (MKSI), formerly Newport Spectra-Physics. The market of ultra-high power commercial lasers has been left to the Europeans, specifically two French companies Amplitude Technologies and Thales (yes, the defense giant). However, NIF and other US national labs still hold significant experience in very large scale lasers and laser optical systems (like how to manufacture them). So the big question is now will this NIF achievement trigger any development in the US laser industry or will they leave the field to the Europeans? My guess is US companies won't pick up the tech because the commercial opportunity of laser-based fusion is still far away and relatively uncertain and the business would be quite 'lumpy', i.e. one could only expect to sell maybe a handful systems per year. I also share your excitement regarding the other technologies, although I think that photonic chips are way too physically large to compete with silicon chips. But the article and paper you mentioned looks really interesting in as they have taken a different approach to training neural networks, not just tried to replicate silicon based algorithms. Hopefully the first CRISPR-based gene editing treatment will be approved by FDA next year - so far it's looking good and on track - first patients that participated in the clinical trials have already been cured.
    4
  2. 27:00 Re smart people make big mistakes - that made me think about Elon admitting that he attempted to automate too much during the Model 3 ramp. Probably the biggest mistake anyone at Tesla ever made and he admitted himself publicly he almost BK the company.
    4
  3. I think their first problem on Mars - before any human sets foot there will be to get a propellant plant going - producing methane and oxygen. That in turn needs a beefy power source (I think ~100kW). According to Mars Direct plan the first trip would deploy such power and propellant plant and start producing propellant for the return trip - so before first humans depart to Mars there is already propellant for return trip ready.
    3
  4. Good (and somewhat surprising) to see Ford getting their EV act togehter with the F-150 and Mach-E. Let's hope they can make those in relevant numbers.
    3
  5. South Korea - not many people have that on their traveling short list. What inspired your interest in SK or have you already been there? (oh and please post your traveling videos)
    2
  6. Yes maybe, but then why did SNAP guide so high? Certainly they must also be aware of the lapping effect and their own past revenues LOL. Anyway, you could likely provide add value if you let us know how your YT income is comparing to past year. Is GOOGL ad business slowing down as well?
    2
  7. Working with like minded people is worth so much! It's part of what makes you want to go to work and also pull you through down times.
    2
  8. What about proper braking? Half the fun of driving fast on curvy roads is in the braking. Most people brake way too timid.
    2
  9. Nice work Nikola!
    1
  10. Amazing
    1
  11. Sam Altman: Intredasting. Lemme clone this repo...
    1
  12. You could feed the Esperato output to Google translate to check ;-)
    1
  13. At first I thought those extra lights are for illumination of the scene.
    1
  14. Would be interesting to apply that to writing the abstract of a scientific article. That is sometimes surprisingly difficult, as it forces you to leave out detail. I image a GPT tool that accepts a paper as input and offers a set of 3 - 5 abstracts to chose from.
    1
  15. David: humans are dangerous, but each individual is limited! Physical limits.... Putin: exists
    1
  16. Hmm, what about this: even if you figure out alignment (let's assume David Shapiro's 3 COF is a solution) - there remains the problem, how do we ensure (enforce?) it's implementation in all A(G)I? Why would malevolent dictator X implement his military AGI with Shapiro-COFs instead of his own COFs? If alignment is actually prompt engineering then it would be fairly easy to manipulate. OTOH, how much "damage" could one (or a handful) of AGI agents do against 8 billion human level intelligences?
    1
  17. I could see a commercial use in marketing: churning out social media posts or blog posts. So basically creative writing in a business context.
    1