Musk is building SkyNet

[Nefertiti]

https://fortune.com/2024/01/23/elon-musk-tesla-fsdbeta-v12-chatgpt-moment-ai/

[Nefertiti]

https://fortune.com/2024/01/23/elon-musk-tesla-fsdbeta-v12-chatgpt-moment-ai/

Of course Tesla Autopilot "ChatGPT moment" is nonsense. On this FSD Musk really sells air. Computer vision (CV) has empirically not hit milestone akin to Natural Language Processing (NLP).

NLP hit big milestone when transformers appeared in 2017. Open AI GPT, Google BERT, Facebook Llama, Baidu ERNIE - all these large language models (LLMs) were enabled by the discovery of transformers a few years ago. Speech and language tech leapfrogged.

But then you have Vision Transformers (ViT) - which significantly improved RNNs for video processing. But not enough for a leapfrog.

There are 2 paths in self-driving research now -
1. Google Waymo uses LiDAR plus cameras - which is great but costly hardware. LiDAR is able to beat bad weather like mist or snow - which even 4k cameras cannot. But LiDAR is expensive and cannot scale - cause you have LiDAR installed on each car - so selling to auto market won't cut it.

Additionally Waymo uses HD Maps - which of course are software - scalable - and Google of course has hitech maps of every corner on the globe. The con of maps is you have to train the CV models on every locality - which makes HD Maps sort of unscalable or hard-coded.

So unscalable LiDAR plus semi-scalable HD Maps - Waymo operates real robotaxis only in Phoenix-Arizona, SF Bay Area and I think Albuquerque-New Mexico. But they have very few accidents.. which is much better than Tesla.

2. Tesla uses only cameras - which are way cheaper than LiDAR - plus they don't use HD Maps. In short Autopilot is pure computer vision - which is very scalable - just cameras and software which is pushable over-the-air. They have deep neural networks which train continuously from every Tesla car on the roads.

Autopilot is perfectly scalable pure software - BUT without the benefit of LiDAR and HD Maps. Tesla cars crash ovyo ovyo in misty or such unforeseeable traffic happenings.

Tesla "FSD Beta v14" - is basically Level 2 autonomy - which is just glide smoothly on freeway - but bonk furiously at every junction or snow or town for driver to take over. If there is fallen tree on the road... or broken down truck - driver must take over or the car parks on roadside 😀😂

Basically Musk promise of "ChatGPT moment" is hot air. Nothing big has happened in computer vision to leapfrog like NLP as far as I know.

Yeah - self-driving "ChatGPT moment" is way off. And will only come when RESEARCHERS  come up with a transformer equivalent for computer vision.

[Nefertiti]

Tesla Autopilot is Level 2 autonomy but cheap. Google Waymo is Level 4 I think... but unscalable and costly.

In short you cannot draw analogy from ChatGPT and other LLMs - which are pure software running on clouds. It will equally take a while for ChatGPT to be available off-the-shelf on ear buds - cause you need the mega-processing power of cloud.

Google AI crew are scratching their heads to release MobileNetV4 with Bard or ChatGPT power. Ngumu - MobileNet is optimized (or "quantized") for handsets which have no GCP or Azure power. ChatGPT runs on 800 petaflops 👀👀 - impossible on mobile.

Maybe Musk has a bridge to sell.

[Nefertiti]

This thread was very short-sighted.. you really do the real George Soros a disfavor 

Jewish oligarchs are real. Dirty Zionist pigs but solid thrifters.

Grok is having ChatGPT and Bard for breakfast, lunch & dinner. No contest. It comes down to access to data:


X as global town-hall is more real-time

Bard has mountains of search data and obviously 🙂 Gmail conversations

ChatGPT has Bing search but a mere subset of Google data


The question is who has the BEST DATA. Grok wins hands down

xAI is worth $20B wants to raise $6B new capital. So far raised only $135m

OpenAI is worth $86B wants to raise $8B. So far raised $13B from Microsoft



Jan 26, 2024
https://www.ft.com/content/28983bdc-2a38-4103-beae-08d9542ab69d

[KenyanPlato]

I couldn't find information about Grok having economic potential. However, ChatGPT and similar AI models have vast potential for economic impact. They can revolutionize industries like customer service, data analysis, and content creation. For more information, you might want to try searching online, as I didn't find specific data on Grok.
https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier

[Nefertiti]

If RV Punda the local charlatan had a working braincell, he'd have retained his AI startup for a few years as AI peaks. At least sell out abit later like OpenAI. Look at the meteoric rise of Nvidia into Trillion-dollar club of five: Microsoft, Apple, Alphabet, Amazon, Nvidia.

Meta and Berkshire Hathaway are within striking distance:- Meta has some skin in the game with FB, IG, WhatsApp data - and their AI team has Torch, PyTorch, AngularJS, etc - notable AI platforms. And of course Llama LLM which is serious ChatGTP and Bard rival.

Berkshire Hathaway has some significant stake in BYD... which is running circles around Tesla in China, APAC and Europe. And soon BYD will take over Africa and Global South. Because they have cracked battery tech and are vertically integrated with a chokehold on global battery supply chain.. which enable them to clobber Tesla on price.

Yeah - Berkshire is near Trillion-dollar club cause stake in BYD, Apple.

But how come Tesla is worth 7- or 10X as much as BYD?

Well, BYD may be able to mint cheap EVs by owning the battery-auto supply chain. But Tesla has Energy game too - with mega-packs feeding the grid in California, Texas, Australia, etc - which feed the Tesla superchargers for their EVs and even Ford or Uber EVs - complete with home- installable "power-walls" or solar panels. Tesla with EVs, superchargers, grid-supplying megapacks and the panels - is in fact a "smart grid" or IoT ecosystem.

Hold on. Musk has Twitter/X data and xAI with just Grok LLM for now. Tesla is also working on Optimus - a humanoid robot. It will improve rather quickly with vision capabilities from Autopilot and language capabilities from xAI.

Hold on. SpaceX Starlink has global constellation of LEO satellites. Which connects all that stuff globally.

Hold on. Neuralink just released "Telepathy" brain-chip. The objective is a pluggable earbud that can connect humans, swines, gadgets, cars, robots, homes, etc to the SkyNet grid.

Did I mention the Boring Company? A "Road OS" tunnel network under the city.

So the kicker is that "Tesla Mobile" - the near-future Musk has been selling of Level 5 cars on Uber-esque network - is buoyed by the smart IoT ecosystem - a complete circle that noone else can muster without big multi-company partnerships.

This justifies the Tesla 5- or 10X vs Toyota or VW or BYD valuation. And the Regular RV Pundit or Georgesoros cannot see beyond their nose. So they insist that Tesla is "waayyy overlued". Do you really think the market is that stupid?

Tesla seems to be struggling with Autopilot or get outpriced by BYD. But personally I foresee it hitting 1, 2- 3Trillion-dollar club. If it doesn't go private soon.

[Nefertiti]

Grok is in alpha release and only available to X premium subscribers. Premium is only a dollar you can get on there and see for yourself.

And why do you need analysis on Grok. It is the same as GPT, Bard or other LLMs. Same business model.

[Nefertiti]

Here https://nipate.net/index.php?topic=12443.60 - Musk is building SkyNet the dystopian "Singularity" in scifi.

[Georgesoros]

His 56bil package was thrown out by the judge.
Like Trump am sure he will appeal all the way to Supremacy
Does this mean he loses his ultra wealth?

[Nefertiti]

His 56bil package was thrown out by the judge.
Like Trump am sure he will appeal all the way to Supremacy
Does this mean he loses his ultra wealth?

Dots & crosses. Musk is major shareholder of Tesla, SpaceX etc - it's robbing the right hand with the left.

[Nefertiti]

Georgesoros Musk has the best performance metric ever: market cap of the company!

Most CEOs are paid to deliver revenues - which of course is suboptimal. Cause you can mint revenues with skeletal margins so zero profits or dismal market share or market cap. Market cap is the ultimate metric to measure performance.

If I am chairman of Vodacom or Safaricom - I will give Tusker dwarf market cap target of $20B and give him all the powers or support he needs - expand abroad, buy startups, launch new stuff, hire foreigners or consultants, increase salaries and the shebang.

All that matters is market cap. The market is smart.

[Nefertiti]

His 56bil package was thrown out by the judge.
Like Trump am sure he will appeal all the way to Supremacy
Does this mean he loses his ultra wealth?

Should Tesla change its state of incorporation to Texas, home of its physical headquarters?

— Elon Musk (@elonmusk) January 31, 2024

?t=86Xs8_cm44OOehja9U28gQ&s=19

[Georgesoros]

The man is a narcisist - "mirror mirror on the wall, who is better than me"?? Every morning.
Investors want return on onvestment, and if all money goes to CEO then what is the point.??

[Nefertiti]

The man is a narcisist - "mirror mirror on the wall, who is better than me"?? Every morning.
Investors want return on onvestment, and if all money goes to CEO then what is the point.??

Musk EARNED that money. In 2019 Tesla was worth $50B with $20B revenues and zero profit. The target was $200B value, $50B revenue, with a positive margin. He blew all the targets.

[Nefertiti]

That dimwitted shareholder should be ashamed of himself. What is the suit about if not cheap publicity.

[Nefertiti]

Country by robot density ie how digitized or automated is the economy.

https://ifr.org/ifr-press-releases/news/global-robotics-race-korea-singapore-and-germany-in-the-lead

[Nefertiti]

Google is spoiling for war 👀

Google debuts subscription Gemini AI bot, as Microsoft rivalry heats up
https://finance.yahoo.com/news/google-debuts-subscription-gemini-ai-bot-as-microsoft-rivalry-heats-up-130006991.html

An important point is that Google Bard (now Gemini) is only an LLM or chatbot without vision features. It is also poorly integrated with their search engine.

Unlike Microsoft Bing search engine strong integration with both OpenAI's ChatGPT and DALL-E which makes Bing a wholesome AI experience including visual or image generation.

[Nefertiti]

Google vs Microsoft...

the former is better for mobile use, and in corollary has more retail audience and consumer products. But in gaming and AI this grip is loosening, peeling away to Microsoft, Meta, AWS. AI developers, for instance, have to contend with the poor architecture of Keras AI platform being haphazardly overlaid on TensorFlow. I often run into the headache of calling Keras via TensorFlow or directly through its own API.

These critcisms are many in Geekdom... StackOverflow, GitHub, Kaggle (this last which Google owns 🙂), etc.

Naturally, I think, Google should soon buy Unity or Unreal... or one of the other popular gaming platforms - as their answer to Microsoft acquisition of Activision Blizzard. Their desperation is nauseating.

Strategically, I see Sundar Pichai as a jarhead drone COO-type ala Tim Cook or Satya Nadella, great at logistics but poor at innovation. This is disappointing because Gates is still power behind the throne , can't say the same for Larry Page, Sergey Brin.

[Nefertiti]

https://techcrunch.com/2024/02/11/googles-and-microsofts-chatbots-are-making-up-super-bowl-stats/

[Nefertiti]

Andrej Karpathy leaves OpenAI. I won't be surprised if he shows up at xAI working on Grok.

Hi everyone yes, I left OpenAI yesterday. First of all nothing "happened" and it’s not a result of any particular event, issue or drama (but please keep the conspiracy theories coming as they are highly entertaining :)). Actually, being at OpenAI over the last ~year has been…

— Andrej Karpathy (@karpathy) February 14, 2024

?t=OGkmbRLed7wxYtJNz2KWkA&s=19