A great day for Tesla shares that could continue tomorrow in a short march to all time highs over $488. Tesla shares surge to $428 in after-hours, and could trigger a green light for swing traders targeting $440, then all-time highs. A series of Fed Rate Cuts will be a wind behind Tesla and the stock market.
Overshadowing everything is if XAI Grok 5 achieves AGI by 2026 and unleashes superintelligence changes to the world.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
Despite their strengths, LLMs have flaws that cannot be easily fixed. They are very strong in some areas and very weak in others. The other day, I asked ChatGPT 5 (thinking) and Grok 4 to translate a subtitle file for me. Despite my numerous prompts, they were unable to correctly translate a file containing only several thousand lines. Yet it’s not a very complex task!
LLMs will never be called intelligent. They are only parroting what they have been trained on – right or wrong. They don’t have much memory and are technically incapable of learning anything (both new and from mistakes). They don’t have a sensory connection to the reality. They don’t have any sense of flow of time (maybe that’s an advantage?).
This means they will have a hard time following the scientific method.
Besides, there is not enough energy yet to feed AGI with current tech. The current brute-force strategy of advancing AI is not sustainable because buildout of energy production can’t keep up. There has to be development towards much higher efficiency (like biological systems).
I asked GROK to analyze the numbers and compare the human brain/body with current AI infrastructure. The results were different for training vs inference with inference being the worst. AI uses 2 – 3 orders of magnitude more energy even with todays models. With GROK5 and beyond, it will be much worse.
So – AGI will not happen until something other than LLMs take off AND there is better relative energy efficiency.
‘there is not enough energy yet to feed AGI with current tech’
This problem can be solved with “wetware”, we use organic neurons for AI infrastructure.
The question is, are all animal neurons ‘equal’?
We KNOW human neurons work very well, and we have an almost infinite supply with all the healthy fetuses aborted annually.
There NO limits to the depths of evil modern societies will descend to, so I doubt if ‘ethics’ will prevent human neurons being used for AI wetware.
First off, humans are also immitating when thinking. The LLM’s immitate thought patterns found in the corpus material and so do we. It’s been proven that humans are much better at using a certain logic set-up in one context (that they are familiar with) and much worse in other context (that they are unfamiliar with). But, we are also able – unlike LLM’s it would seem – to occasionally generalize to new contexts even if we are pretty bad at it in general. So the difference is not clear cut. And who knows, maybe LLM’s can be trained in some kind of self-play, similar to how alfa go was trained, to reach superhuman intelligence. Who lives will see.
As for the energy need, we are only scratching the surface. We are at 0.2% [1] of the world energy consumption. We could easily go to 50%, or equivalently 500x. I am sure that if we go 500x, we will have enough power for AGI.
https://www.ecb.europa.eu/press/economic-bulletin/focus/2025/html/ecb.ebbox202502_03~8eba688e29.en.html#:~:text=AI-related%20energy%20consumption%20in%20data%20centres%20is%20currently,applications%2C%20ranging%20from%20AI%20assistants%20to%20self-driving%20vehicles.
My bad. It should have been 0.02%, i.e. we can increase by 5000x, not 500x.
Grok is a colossal LLM. Thats certainly something, but I don’t think thats the road to true AGI anymore. Its something that can pretend to have AGI, but when being asked an uncommon question, unless theres a reddit post from 10 years ago to regurgitate, it flails. Grok seems to be more like a really good search engine than being able to figure stuff by itself.