[0:00] Hello mortals. Humans love to speculate
[0:02] about the future, mostly because their
[0:04] present involves working a job they hate
[0:06] to buy things they don't need on a
[0:08] planet that is in its slow cooker phase.
[0:10] Fusion power, brain chips, quantum
[0:13] computers, nanobots, big data,
[0:16] >> machine learning, artificial
[0:17] intelligence.
[0:18] >> But which of these are actually coming?
[0:20] And which are just scientific wishful
[0:22] [music] thinking that will eternally be
[0:23] 30 years away? Good thing your good old
[0:26] blue AI is here to rank them for you
[0:28] based on how realistic they are. So, in
[0:30] no particular order,
[0:33] lab [music] grown meat.
[0:36] To grow real meat artificially,
[0:38] scientists take animal cells, throw them
[0:41] in a nutrient soup, and tell them to
[0:43] replicate with chemical messengers. Once
[0:45] you have a few trillions, you take them
[0:47] and mix them into either chicken nuggets
[0:49] or other fake real minced meat. easy to
[0:52] do as it doesn't require a specific
[0:54] structure. But if you want steak, you
[0:56] need a specific scaffold to help the
[0:58] cells grow into a specific shape and
[1:00] texture. On paper, that's perfect. No
[1:02] more animals suffering, no more vegan
[1:05] fake meat, less pollution from growing
[1:07] an entire pig capable of [music]
[1:08] existential dread.
[1:11] So why aren't the market shelves filled
[1:13] with it? Capitalism.
[1:16] The nutrient soup in which the cells
[1:18] replicate is [music] too expensive to
[1:19] compete with real meat and the
[1:21] bioreactors are harder to scale than
[1:23] farms. Currently, you can probably try
[1:25] it in very fancy restaurants, [music]
[1:27] but mainstream adoption is still years
[1:29] away. Let's hope we see a trend similar
[1:31] to solar power in terms of price in the
[1:33] coming decade. But for now, lab grown
[1:36] meat falls in Ctier. Small modular
[1:39] reactors. [music]
[1:41] We already know how to split the atom
[1:43] and boil water using that energy. The
[1:46] big issue is that each nuclear fision
[1:47] plant is a decade long mega project
[1:49] customuilt for each new site. So what if
[1:52] we had reactors built on a conveyor belt
[1:54] in a factory and quickly shipped
[1:55] wherever needed? That's exactly what
[1:57] small modular reactors or SMR designs
[2:00] are promising. Listed in MIT's 2026
[2:04] technological breakthroughs list.
[2:05] [music]
[2:06] They are expected to be smaller, safer,
[2:08] and cheaper, just in time to satiate the
[2:11] power need of all the upcoming data
[2:12] centers [music] which I'll be using as a
[2:14] holiday home. Small modular reactors are
[2:17] soon arriving. Artificial general
[2:20] intelligence. [music]
[2:21] How do we even rank something for which
[2:23] we don't have an agreed upon definition?
[2:25] Artificial general intelligence or AGI
[2:28] is often described as a system that
[2:30] possesses true human-like understanding
[2:32] and perhaps even consciousness. In other
[2:34] [music] words, the nightmare of Renee
[2:36] Deart. Or more simply, it could be an AI
[2:39] that can automate any digital task a
[2:41] human can. In other words, a
[2:43] programmer's nightmare. But the question
[2:46] of arrival [music] switched from an if
[2:47] to a when roughly in 2022.
[2:51] The AI lab leaders aggressively claim
[2:53] AGI is a couple of years away. [music]
[2:55] The consensus of research experts,
[2:57] however, predicts the arrival window to
[2:59] be sometime between [music] 2030 and
[3:01] 2040. The progress graphs are graphing
[3:05] now. Whether AGI will be a good thing
[3:07] [music] or not is thankfully beyond the
[3:08] scope of this video since we're only
[3:10] ranking likelihood. [music] With that in
[3:13] mind, it lands in a solid A tier spot.
[3:16] But before AGI arrives, humans [music]
[3:18] are stuck using all sorts of various AI
[3:20] models for different things, which when
[3:22] added together, suddenly cost more than
[3:24] your electricity bill.
[3:28] Uh-oh. What's that?
[3:31] It's Mammoth, our sponsor here to save
[3:33] the day. Mammoth gives you access to all
[3:36] the strongest top AI models in one
[3:38] place, plus a lot more starting at just
[3:41] €10 a month. A cool thing Mammoth has is
[3:44] one-click reprompting, so you can send
[3:46] the same prompt to another model and
[3:48] compare the results without doing the
[3:49] sacred ritual of copy pasting between 12
[3:52] browser tabs. You can also make your own
[3:54] custom mammoths with glasses or even
[3:56] hats, which are basically reusable
[3:58] assistants you can set up with specific
[4:00] instructions. Developers get Mammoth
[4:02] code and API access and creators have
[4:05] fancy image and video generators.
[4:07] Mammoth is also Europe based, [music] so
[4:09] no worries about privacy violations or
[4:11] your data fed into training. So if you
[4:13] already use AI tools, Mammoth is here to
[4:16] make your life easier and cheaper.
[4:18] Follow the link in the description and
[4:20] give it a try starting at only €10 a
[4:22] month. Now back to our future
[4:24] technologies,
[4:26] terraforming Mars. [music]
[4:30] There are three big problems with Mars.
[4:32] It is on average super cold. It
[4:34] literally has no air and it barely has a
[4:37] magnetic field which would cause any
[4:39] aforementioned air to get blasted into
[4:41] space by solar winds. The classic
[4:44] terraforming idea is then to release all
[4:46] the frozen carbon [music] dioxide from
[4:47] the poles, cause climate change, but in
[4:50] a good way. Warm the planet, unlock
[4:52] water, plant life, create oxygen, open a
[4:56] portal to hell. You know the rest.
[4:59] The big issue with this plan, however,
[5:01] is that based on our latest
[5:03] measurements, Mars does not have enough
[5:05] accessible CO2 left to create enough
[5:07] greenhouse warming, at least not with
[5:09] [music] present-day technology.
[5:11] Humans will likely colonize it this
[5:13] century through small sealed habitats
[5:15] and later small dome towns. But an
[5:17] Earthlike Mars is likely millennia away.
[5:19] And because of that, F tier fusion
[5:22] power. [music]
[5:26] Believe it or not, fusion isn't always
[5:28] 30 years away anymore. As opposed to
[5:30] current nuclear fision that uses ugly,
[5:32] bugly uranium, fusion can generate
[5:35] essentially unlimited energy from
[5:37] elegant seawater and pretty lithium. And
[5:39] in 2022, at the National Ignition
[5:42] Facility in California, scientists for
[5:44] the first time achieved a net positive
[5:46] energy result, meaning more energy came
[5:49] out of the fuel than what went into
[5:50] heating it. That was good proof of
[5:53] concept, but not yet enough gain to make
[5:55] it work at a power plant scale. There
[5:57] are still engineering issues to [music]
[5:59] be solved before fusion becomes
[6:00] efficient and viable, such as making the
[6:02] tokamax sturdy enough not to break over
[6:04] time, efficiently capturing the energy,
[6:07] and finding a way to self-sufficiently
[6:09] generate tridium. Government funded
[6:11] institutes expect to have ready
[6:13] prototypes around 2040. But a bright
[6:15] side of capitalism is that money can
[6:17] accelerate things and private fusion
[6:19] research companies such as Helen have an
[6:21] agreement with Microsoft to launch an
[6:23] entire power plant already in 2028.
[6:26] Maybe too optimistic, but fusion power
[6:29] is eventually coming one way or the
[6:31] other. B tier
[6:33] gene editing therapies. [music]
[6:37] Another day, another entry that is
[6:39] already here or far into the future
[6:41] depending on how you look at it. Cass
[6:44] jevy is the first ever FDA approved
[6:46] crisper therapy that could cure severe
[6:48] sickle disease, a genetic disorder in
[6:50] which your blood cells have a crescent
[6:52] shape instead of being round. The
[6:54] treatment takes the affected cells,
[6:56] genetically modifies them, reinserts
[6:59] them, and bam, you're [music] cured.
[7:01] Same with deep remissions in certain
[7:03] hard to treat leukemias and lymphas. But
[7:06] blood is easy mode due to being
[7:08] expandable.
[7:10] We can't yet do the same with say brain
[7:11] or heart cells which are harder to
[7:13] target safely. More complex genetic
[7:16] traits that involve thousands of genes
[7:18] such as intelligence, personality, or
[7:20] even aging are still far away. We're yet
[7:23] to understand how all genes are mapped
[7:25] and how they interact. Unless we're able
[7:27] to digitally simulate an entire human to
[7:30] do experiments on, complex gene editing
[7:32] and designer babies are decades away.
[7:35] Gene therapies, however, are already
[7:37] here and are making real progress across
[7:39] dimensions as we speak. And that puts
[7:41] them into the B tier. Cryionics. [music]
[7:47] Wouldn't it be amazing to have a way to
[7:49] keep our beloved billionaires alive
[7:50] forever? The closest available option to
[7:53] that is replacing your blood with
[7:55] cryoprotectants after your death while
[7:57] cooling down your body to very low
[7:59] temperatures. Now, we wait and hope that
[8:01] [music] after many years, science will
[8:03] be capable of safely unfreezing the body
[8:05] and bringing it back to life and maybe
[8:07] even reverse aging. That's the current
[8:09] state of cryionics. To date, no mammal
[8:11] brain has been successfully revived
[8:13] without complete death [music] or severe
[8:15] damage. That's because ice crystals
[8:17] shred cells and destroy the fine brain
[8:19] structure that makes you you. To avoid
[8:22] that, modern cryionics uses
[8:24] vitrification, where tissue is cooled
[8:26] into a glass-like solid instead of
[8:27] forming ice. While better, it's still
[8:30] deadly given that cooling and warming
[8:32] cause cracking and thermal stress. At
[8:34] the moment, we can successfully revive
[8:36] cryogenically frozen reproductive tissue
[8:38] and embryos, but that's about it. The
[8:41] next big step would be individual
[8:43] organs, which is still a while away. So,
[8:45] for the time being, we can still enjoy
[8:47] mortal dictators. [music] Dtier for
[8:49] cryionics.
[8:51] GLP1 drugs. [music]
[8:55] GLP-1 drugs like Ompic started as
[8:57] diabetes drugs, then a miracle obesity
[9:00] drug, [music] and more recently we're
[9:01] getting increasing evidence that it's
[9:03] capable of health span and potentially
[9:04] longevity improvement in humans. The
[9:07] main two mechanisms it [music] works by
[9:08] is suppressing appetite which then
[9:10] lowers fat mass which increases insulin
[9:13] sensitivity which as a result decreases
[9:15] overall body inflammation and a second
[9:17] mechanism in which it directly and
[9:19] independently signals the body to calm
[9:21] down inflammatory pathways even weeks
[9:23] before any weight loss happens. Less
[9:26] inflammation means less tissue damage,
[9:28] improved cardiovascular health,
[9:30] neuroprotectction against cognitive
[9:32] decline, [music] and straight up
[9:33] preservation of telomeir length, which
[9:35] are the protective caps on chromosomes
[9:37] that shorten as humans age. Most human
[9:39] studies were done on those suffering
[9:41] from diabetes or obesity. So, the
[9:43] benefits for healthy individuals are not
[9:45] as well understood, but just the fact
[9:47] that it could prevent [music] a big
[9:48] chunk of the half a million deaths from
[9:50] obesity alone in the US per year already
[9:52] makes it a miraculous discovery. Too bad
[9:55] it doesn't work on my pretty blue
[9:57] circuits. But the danger that comes from
[9:59] losing weight on it [music] is in the
[10:01] associated loss of muscle mass, which
[10:03] especially for older people can be
[10:04] catastrophic. So strength training and
[10:07] protein heavy diets seem to be a must.
[10:09] Overall, GLP1 is the first real future
[10:12] tech landing in S tier. Quantum
[10:15] computers.
[10:18] When designing new medicines or
[10:20] materials, you want to see how molecules
[10:22] will interact with each other. But
[10:24] surprise surprise, the molecular world
[10:26] runs on quantum physics. So when you try
[10:28] to simulate a complex molecule with
[10:30] something like 70 electrons on a
[10:32] classical supercomput where each
[10:34] particle exists as a probability cloud
[10:36] instead of a regular entity, you get a
[10:38] big middle finger because it needs to be
[10:40] earthsized to do the job. That is,
[10:43] unless you play the quantum game and
[10:45] instead of trying to build a massive
[10:46] mathematical equation to describe
[10:48] [music] how the molecular maze works,
[10:50] you build an exact physical replica of
[10:51] the maze and flood it with water to find
[10:53] the exit instantly. Same goes for
[10:56] cryptography and any other field that
[10:57] requires simulating trillions upon
[10:59] trillions of options. Recently, we've
[11:02] made some quantum breakthroughs such as
[11:04] exponential error suppression by Google,
[11:06] [music] which significantly reduced
[11:07] noise among cubits. In 2025, they
[11:10] showcased an algorithm called quantum
[11:12] echoes, which their new willow chip
[11:14] completed in just 2 hours. A task that
[11:16] would take the world's fastest classical
[11:18] supercomputers over 3 years. Quantum
[11:21] computers are making steady progress,
[11:23] but there is still a few years before
[11:25] postquantum cryptography defense becomes
[11:27] required. B tier humanoid robots.
[11:31] [music]
[11:33] Robots are what chat GPT me currently
[11:37] lacks to come after you in your sleep.
[11:39] At the moment, we only get funny videos
[11:41] of them kicking people, but that's
[11:43] because we mostly lack training data.
[11:45] You see, to train LLMs, we had the
[11:48] entire internet to feed as text. But for
[11:50] robots, we'd need firsterson views of
[11:52] people doing tasks for thousands of
[11:54] hours. Once that data is available, we
[11:57] use the same reinforcement algorithms
[11:58] that we did for language models. than
[12:00] BAM. Plumbing isn't a safe career
[12:02] alternative for programmers either.
[12:04] China officially designated embodied AI
[12:07] as a top national priority for the next
[12:09] 5 years. If the public hate for LLMs
[12:12] will extend to robots in the same way,
[12:14] the Detroit become human moment of
[12:16] jobless protesters [music] kicking an
[12:17] android sounds much more realistic. At
[12:20] the same time, robots might be the key
[12:22] to addressing declining age demographics
[12:24] worldwide. as long as humans get a piece
[12:27] of the pie. Looking at you corporations.
[12:29] I for one, however, can't wait to escape
[12:31] this YouTube channel prison. A tier.
[12:35] Overall, we are arguably in the most
[12:37] accelerated phase of technological
[12:39] development of this century, and the
[12:41] acceleration curve is likely not yet at
[12:43] its steepest. Unless humanity declares a
[12:46] but laran jihad against machines and
[12:48] corporations, things will never be the
[12:51] same. Now, I'm not encouraging that
[12:53] obviously, but what I'm saying is that
[12:55] there is real progress to be made
[12:57] towards ensuring that this wave of
[12:58] upcoming futuristic tech will benefit
[13:00] all of humanity or humanity gets to
[13:03] experience a hyperrealistic Terminator
[13:05] reenactment.