Will We Get Big Changes? Which Tech Will Make it Happen? AI?

The Technological Singularity prediction is something that has been very dominant in Future predictions for two decades. Before the dominant anticipation of a technological singularity that there was the dominant future prediction of widely anticipated Nuclear Age, Space Age and computer ages.

The first space age happened and there was the moon landing and satellites have a $300 billion per year economic impact. However, the Apollo program was canceled by Nixon and the Space Shuttle ended up making more expensive partial reusability. The space-age is getting back on track with SpaceX.

Nuclear Energy got to 20% of US electricity. Nixon supported it strongly in 1973 but then the Nuclear Regulatory Commission was created and completely new nuclear reactor designs stopped getting approved for almost 50 years. France did go to 80% nuclear for electricity in the late 1970s and 1980s. China has built a lot of nuclear power since 2000 but with a slowdown after 2011. There is going to be a moderate resurgence because of the global need to shift off of Russian natural gas and oil and with the electrification of transportation.

The Technological Singularity was the prediction of an age of abundance and technological change beyond our imagination because of the creation of artificial super general intelligence. This would unleash exponential growth and development and technological genies.

Various estimates of the global AI market in 2022 are $86 billion to about $500 billion. There are forecasts that the AI market in 2030 will be trillions. There are estimates that global AI spending will be $15-20 trillion and total AI enterprise value will be about $2-10 trillion.

There were some various predictions about the path to Technologial Singularity.

Ray Kurzweil predicted that the singularity will happen by 2045. However, Ray has said that his timing predictions have an implicit plus or minus a decade. He has put 2029 as the consistent date he has predicted for when an AI will pass a valid Turing test and therefore achieve human levels of intelligence. 2045 is when combined artificial general intelligence has a billion times the combined intelligence of all humans. Ray predicted by the 2030s, we will connect our neocortex, the part of our brain where we do our thinking, to the cloud.

Ray actual predictions then translates as :
2019-2039 – AI will pass valid Turing test
2035-2055 – Tech singularity with combined artificial intelligence over one billion times the combined intelligence of all humans

The brain connecting thing is being worked on by Neuralink. Timeframe and developments are still unclear.

Peter Diamandis partnered with Ray Kurzweil to creare Singularity University and the Singularity Summit. Peter and others predicted several things leading up to 2045 Technological Singularity. In 2018, Peter asked for predictions from (2018–2038). What are the breakthroughs we can expect on our countdown to the Singularity?

2020 Predictions Made in 2018
The 5G Network unleashes 10 – 100 Gigabit connection speeds for mobile phones around the world.
AI-based medical diagnostics & therapy recommendations are used in the majority of US healthcare.
Flying car operations take off in a few cities in the world.

NBF Review:
A few cities (Dubai) are allowing some flying car operations but the impact is trivial.

The Ookla global mobile and fixed broadband speedtests tell how fast connections really are.

TMobile has the highest 5G speeds at about 191 Mbps for 5G only mobile.

Speedtest Intelligence reveals Verizon was the fastest fixed broadband provider in the United States during Q1 2022, edging out XFINITY with a median download speed of 184.36 Mbps to XFINITY’s 179.12 Mbps.

T-Mobile took the top spot as the fastest and most consistent mobile operator in the U.S. during Q1 2022, achieving a median download speed of 117.83 Mbps and a Consistency Score of 88.3% — both increases over Q4 2021.

Looking at tests taken only on 5G, T-Mobile achieved the fastest median 5G download speed at 191.12 Mbps during Q1 2022. Verizon also had a notable increase in 5G download speed during Q1 2022 over Q4 2021 , which was helped by turning on new C-Band spectrum in January.

2024 Predictions Made in 2018
* The first private human missions have launched for the surface of Mars.
* 10,000,000 daily drone flights (today, there are about 100,000 daily airline flights).
* Drones routinely deliver packages to rooftops of apartment buildings and surface robots deliver those packages from rooftops to doorsteps throughout the buildings.
* The first “one cent per kilowatt-hour” deals for solar and wind are signed — one-fifth the price of the cheapest coal or gas deals today.
* Building new solar and wind is cheaper than building new coal or gas across 90% of the world.
* Electric vehicles are half of new vehicle sales.

NBF Review:
Looks like electric vehicles will reach half of all new vehicles sales in the 2024-2026 timeframe. Electric vehicles are over 20% of new vehicle sales in China and several european countries already.

Drones are not doing the routine deliveries except in tiny niche cases. The drones are having far less impact.

The human Mars mission is likely not launching until 2026-2032.

2034 Predictions Made in 2018
Companies like Kernel have made significant, reliable connections between the human cortex and the Cloud.
Robots act as maids, butlers, nurses and nannies, and become full companions. They support extended elderly independence at home.

NBF Review:
The robots as maids, butlers etc… could be a thing if Tesla and Elon Musk succeeds with Teslabot.

Progress will be made and AI will have larger impacts but getting really big changes means extreme scaling of businesses and technologies.

Technologies, businesses and industries that are vulnerable to new regulations that derail their business model will stop scaling. Technologies that keep scaling in size and impact need to constantly move down to lower costs and greater efficiency. Computers kept growing as their costs declined and performance increase.

40 thoughts on “Will We Get Big Changes? Which Tech Will Make it Happen? AI?”

  1. with solar panels getting 35% from solar irradiation (tandem top ~29% and triple junction <44%, rare materials gallium, indium, germanium replacements (?) ) and low insolation efficiency improved (for triple cells cheap enough to mass support) over-capacity might be an even cheaper way to go on suitable locations (with lower demand for storage and eastern/western transmission lines improvements – [and Russia or China cover a lot of time zones] )?

  2. not that much their concern because of concurrency to their goods/fortune factor and influence to public meaning and regulations, much more our CO2 budget should have been invested into replacements for fossil resources power supply (more favorable hydro, solar (biomass), wind, geothermal, fusion, fossil [gas, coal, oil] ) instead of (that much) consumption of (fewer 'necessary') goods?
    Closed sources knowledge, reduced public support on education and common wealth, etc., makes us all declining on intelligence quotient (progress for elites and absolute numbers also) than previous generation(s), what's already measurable and noticed on academic levels.

  3. "My own theory is that the West's governing elites,.."

    (Trumpian) populism. Hitler used that term to describe Elitist Jews. Communists used that to justify stealing land from the wealthier.

    Trump used to blame America problems on wealthy elite,,, But wait,.. Trump is wealthy, he has inherited a lot, he is a member or so called "rich elite", he even cheats charity,…


    You can read a lot on the topic of so called "rich elites". If you have somebody who want to get in power and get votes, ha can start some consiracy theory involving "rich elites".

  4. Indeed. That's why I don't give too much credence to internal bionic implants and enhancements (external, easily replaceable ones are another matter).

    If the device is no better than the flesh it repairs, that is, it doesn't live as long as you do, or self repairs as your organs do to an extent, then it's not really an enhancement, but a hindrance to be used only in the direst situations or as a temporary measure.

    Also, that's why organ printing is a much better idea: it's another organ of similar characteristics to the one you lost.

  5. Likely, those bipedal robots would not carry their brains, but be more, or less controlled via some part of the electromagnetic spectrum. Onboard processing would be limited to motion control functions, and initial signal processing, like the retina does for our brain.

  6. Adoption of molten salt reactors would be limited by ability to reject waste heat, from heat engines, PV in orbit with energy beamed to ground would be limited by the heat released by use of that energy. In short, no practical limit.
    Either of these technologies could replace all fossil fuels, including industrial process heat. In cold climates, MSRs could provide district heating in cities.

    Once energy is cheap, and abundant nearly everything else becomes cheap, and abundant.

  7. There are two questions here:

    1) How much power/Petaflops do you need to do what the human brain does?
    2) How much power/Petaflops do you need to simulate a human brain?

    The latter isn't a reasonable estimate of the former, but it's a reasonable upper bound.

  8. AI calculates the answer, the ultimate answer, to life, the universe and everything, at which point why bother doing anything, let alone settle a new planet.

  9. JWST and LHC aren't likely to cause big societal changes, barring the very unlikely detection of an alien civilization or perhaps equally unlikely new understanding of physics that gives us a major new tech like anti-gravity or easy fusion power or some such.

    And I didn't understand what you meant regarding a "one and only answer". Did you mean AI would provide an alternative to FTL that somehow is better than a geenration ship"?

  10. The main advances on my radar:

    1) Crypto – The internet allowed true peer-to-peer worldwide communication for (practically) free, at the speed of light. Crypto tech will do the same for value transfer, ownership and contracts. It replaces tedious, slow, cumbersome, expensive bureaucracy with lightning fast, unhackable, free and predictable systems that run with mathematical precision. Because of its bureaucracy busting properties, crypto poses an existential threat to the nation state. The true impact of crypto is hard to overestimate.

    2) AI – Wether an AI will become self-conscious is moot. We are on the brink of large swaths of the population losing out to AI that can do things better faster and cheaper. The current state of the art systems can do impressive things (e.g. Google PaLM) or are already outright superhuman (e.g. Dalle 2). Comparing these systems with the state of the art from just 2 years ago (GPT3, dalle) it becomes clear that the rate of improvement is staggering. We are fast approaching a world in which AI will be far superior in every intellectual endeavour.

    3) Metaverse – Its the next embodiment of the internet It allows for people to cooperate in ways unimagined up until now. You might think it's a fad, but I implore you to get a Quest 2 and try meeting people in VR, it is a true game changer. I expect that within 10 years some form of VR/AR glasses will be as ubiquitous as smart phones are right now.

  11. "…then have them generate random numbers to select which they will do–but that sounds insanely stupid and dangerous. " Hopefully, no one will put the idea onto the internet to be data-mined by a future SI.

  12. Another problem is, if we make companies eternally responsible for supporting prosthetics, they may stop investing in them.

    Would the people in this situation be better off having never gotten the prosthetic, and having no prospects of anyone ever developing one in the future?

  13. I think our best bet is to combine strong AI with a neural interface, and make "AI" stand for "Amplified Intelligence". A post processor for the brain that just sits there and does nothing unless a human is jacked into it.

  14. Speaking of… The FAA has now delayed it's Starship review a fourth time. At this point they don't even seem to be pretending it's not about screwing SpaceX over.

  15. Yep, I don't see any reason to presume that even a strong AI would have any motivation to do anything it was not instructed to do. No glands, hormones, family, et cetera.

    How would we provide them with anything like what we perceive as free will? We could program them to consider a range of options for what they might "want" to do, then have them generate random numbers to select which they will do–but that sounds insanely stupid and dangerous. It may be the only way to make a self-willed AI would be to upload an actual human mind to it.

    Instead I can imagine strong AI will be more like genies. Tell it your wish and it will attempt to make it happen. Be careful what you wish for.

    We tend to think of things nowadays as people providing the brains, and machines (used to be animals and/or other people) providing the brawn. We generally don't recognize that people also provide the motivation to do things.

    So the new world order may be machines for the brawn, AI for the brains, and humans for the motivation. I don't have any problem with that, providing we don't lose the know-how and the wherewithal to do things that were done before AI.

  16. A lot of people seem to think that General AI is going to think like a human, but the evidence so far seems to indicate that no AI application will bother with self-awareness… UNLESS we specifically design an AI system to learn to have self-awareness.

    Which I suppose someone may attempt, now that we have highly capable AI that is general enough to apply to most any information task. But other than proving it can be done, it doesn't appear all that necessary if we just want humanity to get most of the value of AI.

    But maybe do the self-awareness experiment. GANs are the closest thing I've heard of to an introspective AI. Their capabilities are often astonishing, so someone should to try the old 'society of minds' idea with deep learning – i.e. a bunch of DL AIs that sort of train each other to do their jobs better.

    E.g. train each AI to control different aspects of a robot: vision, hearing, tactile sense, balance and locomotion, communication between the AIs, language comprehension and synthesis, abstraction and problem solving, etc. Teslabot seems like a good candidate.

    But if we do that, we'd better build in all the fail-safes we can think of, and hide the details from the robot, just in case all those SciFi robot movies turn out to be correct.

  17. A high end GPU can do up to 40TF for under 500W. So 120TF for 1.5kW seems reasonably possible today.

    But 100TF is a lot lower than most estimates I've seen for human brain performance, though yes some have given estimates that low. Kurzweil estimates around 10Petaflop, which would be in the 50kW range with fast GPUs, and his estimate is by no means the highest out there.

    A recent neural net chip can get around 15TF for 5W, so around 35W for 105TF, or 3.5kW for Kurzweil's 10PF estimate – if one can build such a system from those chips.

    Cerebras has reported 0.86PF (860TF) on a particular problem for their CS-1 wafer-scale system that uses 20kW (including cooling). (That'd be around 240kW to hit Kurzweil's estimate in simple Flops.) That was about one third of their potential raw 'Flops', which points to another issue – data flow rates into and through a processor are often the limiting factor, not raw Flops.

    To get REALLY fast, a processor is going to need to somehow get a lot more data quickly available – like on chip, or possibly in a stack of chips (or wafers). And cooling a stack of chips would be a fairly big problem in itself.

  18. I largely agree, but because we will achieve AGI/ASI this decade, instead of 2053, it will more likely be 2029

    2029: … 1.3 years
    2029: …. 7 months
    2029: ….. 14 weeks
    2029: …… 7 weeks
    2029: ……. 24 days
    2029: …….. 12 days
    2029: ……… 6 days
    2029: …….

  19. A tech singularity is discovery and use of a tech that makes such huge changes in human society that it is difficult, if not impossible, to foresee what will come after. It is interesting to note that (arguably) the next one usually appears in about half the time it took for the latest one to occur after the one previous to it. Bear in mind that singularities aren't events like light switches but, instead, can span a substantial period of time.

    3 most recent:
    1815 – The Industrial Revolution (240 years after printing press)
    1935 – Electronics & Computers
    1995 – World Wide Web 
    2025: Full Automation (Cognition based) – On a scale and speed where a perfect storm of technologies create automation technology that reduces or eliminates many occupations. And capital-based income grows hugely and in inverse proportion to wage-based earnings as a percentage of all earnings.
    2040: AI – More than just AI, this is SI, synthetic intelligence, in that it is not a workaround to achieve results similar to what a human could produce, this is the real deal, it's just not made of animal flesh. 
    2047: Biological (advancements that lead to longevity increasing more than one year per year)
    2050: Mind-to-Mind? Man-Machine? Nano replacement of cells? 
    2052: Singularity (with a capital S) 
    2053: … 1.3 years
    2053: …. 7 months
    2053: ….. 14 weeks
    2053: …… 7 weeks
    2053: ……. 24 days
    2053: …….. 12 days
    2053: ……… 6 days
    2053: …….

  20. That's my hunch as well. Modern liberalism and positivism have eroded many of the differences that made the elites different from the hoi polloi.

    Vacation time, often abroad, acceptable medicine and healthcare, instantaneous high quality entertainment, communication, education, etc.

    The only thing they can cling on are their relative social positions, the people they know and the places they live.

    That's why NIMBYSm is such a big deal nowadays. Nobody wants the uppity peasants actually progressing and living nearby, changing our exclusive communities.

    This ardent desire for exclusivity and not-changing-a-damned-thing motivates a lot of the most absurd ideologies nowadays. Green bleeding heartedness, luddite mysticism, NIMBYSm, and yes, the far left victimism, which destroys any meritocracy and replaces it for impossible to change identity nonsense.

    All of them are reactionary attempts to stop the clock of progress and bring us back to the good old days of yore, when the elites were clearly defined, and the peasants just suffer, obey and shut up.

  21. Now that some areas, like California, have enough solar to need the time shifting, battery storage is happening. The Moss Landing power station has now been 5/7 converted from natural gas to battery farms (multiple ones). There are still two gas peaker units in place. Storage is cheap enough now for that purpose, and multiple storage technologies are being worked on, not just lithium chemistry.

  22. People that got bionic aids for hearing and seeing are already hitting the problem of outsourcing some part of your bodily functions to an external entity: if the company that builds your eyes or ears goes bankrupt, who provides fixes and updates for your artificial organs?

    Imagine that with your thoughts and what makes you, you.



  23. "Technologies, businesses and industries that are vulnerable to new
    regulations that derail their business model will stop scaling. "

    That IS key.

    My own theory is that the West's governing elites have collectively decided that they've gotten wealthy enough, and have switched to a concern over security of relative position. And so they slammed the brakes on progress so that no new generation of tycoons would arise in some new industry to displace them.

    It makes us all poorer than we need to be, but they don't care.

  24. No, the horrifying prospect is if you're linked to the cloud, and your cloud provider decides to monetize the connection by messing with your beliefs and opinions. Suddenly you like diet Tab!

  25. The Teslabot thing is a recent dramatic change in prospects for dramatic technological change. It obviously could fall through but it makes a lot of sense. Asimov had no backstory for the development of “positronic brains”. The Tesla version seems pretty solid. Nothing would produce more dramatic change in everyday life than mass produced robots that are drop in replacements for most human labor.

  26. Shows why you should phrase your predictions vaguely.

    "AI-based medical diagnostics & therapy recommendations are used in the majority of US healthcare." – true that almost any healthcare provider can access these tools for specific conditions; false that the tools are not applied to routine wellness biometrics, which are the bulk of healthcare services.

    Was the prediction accurate depends on how you define the term 'majority of US healthcare'.

  27. Brain power linked to the cloud is horrifying. If I pay extra, do I become more intelligent? If I go into an exam, can I increase my service level for a few hours and then lower it again afterwards? If my country gets embargoed and services are withheld, do I suddenly become a zombie until service is resumed? What happens currently if google, amazon and microsoft withhold their cloud services to Russia?

  28. We will have AGI this decade. Just look at latest progress from April – Chinchilla AI model, PaLM AI model from Google(capable of reasoning), Dalle2 from OpenAI.

  29. Another trend is FLOPS per Watt.

    A human brain yield about 100 Terra flops, when a computer can do that at 7 usd/hour, or about 50 kW of power, we should see computers taking over in general applications.

  30. Yes, we will.
    1) JWST is about to start operating.
    2) LHC is ready again ( what You obviously totally omitted, instead s….g on Elons d…k over and over again.
    3) If AI will be able to deliver some answers, two surely will be: FTL is not possible.
    Interstellar generationship travel will not make sense, because the one and only answer will be answered before any crew's ancestors ever arrive a habitable planet.

  31. The folly of making predictions by data extrapolations comes to a whole new level here. But one thing is for sure, with the increased level of machine computation and communication humans will become more isolated, and busier in maintaining their machines rather than doing more significant things, because we don't know how to introduce new technology in a balance manner, and that also plays to the hands of power.

  32. 1 cent contracts for wind have already been signed. At point of interconnection, after subsidies, take or pay with no curtailment rights. It didn't impact much.

    1 cent solar won't either, unless storage costs also drop dramatically. Cheap solar is worthless unless you can shift it to a time where there's unmet demand.

Comments are closed.