Top AI Researcher Reveals The Scary Future Of Employment

00:32:04
https://www.youtube.com/watch?v=MCs8BNPYtOA

Sintesi

TLDRThe video presented by Avital Balwit focuses on the transformative impact of AI on future employment. As Chief of Staff at Anthropic, Balwit shares insights into how technological advancements, especially AI, could disrupt the traditional workforce and end many jobs we know today. The discussion suggests that while manual and highly regulated jobs may face slower automation, online and knowledge-based roles are more vulnerable. With this rapid change, there’s a critical need for individuals to adapt by acquiring new skills and considering proactive measures like Universal Basic Income. The psychological effects of unemployment highlight the importance of finding value beyond economic contributions. Balwit underscores the urgency to prepare and adapt as AI continues to evolve, posing both challenges and opportunities for society and the future economy.

Punti di forza

  • 🤖 AI is rapidly changing the landscape of employment.
  • 💼 Many traditional jobs may become obsolete due to AI.
  • 📉 Unemployment might increase leading to social and psychological impacts.
  • 👨‍🏫 Regulated industries like medicine and law may remain secure for longer.
  • 🛠 Manual and physical jobs are less susceptible to immediate automation.
  • 🧠 Finding purpose beyond work is crucial for future happiness.
  • 📊 Universal Basic Income could become vital in securing basic needs.
  • 🌐 Online jobs and remote work are first in line for AI disruption.
  • 🔍 There's a need to understand AI developments to prepare effectively.
  • 💡 Skills adaptation and lifelong learning are key to staying relevant.

Linea temporale

  • 00:00:00 - 00:05:00

    This video emphasizes the significance of insights from Avital Balwit, chief of staff at Anthropic, about the future of employment, particularly under the influence of AI and AGI. It underscores the potentially transformative impact of AI on employment, urging viewers to prepare for this shift by understanding and investing in AI-related fields to stay ahead in a rapidly changing economic landscape.

  • 00:05:00 - 00:10:00

    Avital Balwit, reflecting on her career at Anthropic, shares her perspective on technological advancements and their potential to end traditional employment. Her statement that AI developments might soon obviate her need to work provokes a deeper discussion on the implications of AI advances such as language models for the job market, highlighting growing automation and the diminishing role of human labor.

  • 00:10:00 - 00:15:00

    The video continues by exploring the general denial among knowledge workers about AI's ability to replace human tasks. It discusses AI's automation capabilities, emphasizing the progressive obsolescence of certain skill sets like freelance writing due to AI improvements. The narrative challenges the common underestimation of AI impact, urging a shift in perception to adapt to inevitable economic changes.

  • 00:15:00 - 00:20:00

    It addresses the broader societal implications of AI's ability to perform nearly all economically useful tasks, foreseeing significant automation of online and remote work industries. It lists sectors like copywriting and customer service as key areas facing automation. The importance of understanding AI's potential to replace average human performance in tasks is emphasized, projecting further shifts in labor dynamics.

  • 00:20:00 - 00:25:00

    The discussion moves to exploring how societies will manage employment obsolescence and individual happiness without work, given AI's growing role. It touches on universal basic income as a possible solution to meet material needs and discusses the psychological impact of unemployment, stressing the importance of finding fulfillment beyond traditional labor roles in a future dominated by AI.

  • 00:25:00 - 00:32:04

    Finally, the video covers the potential positive roles of AI in society and how individuals can prepare for a post-AGI world. It highlights the need for humans to find joy in activities beyond economic necessity and addresses concerns over unemployment due to AI advancements. The narrative offers strategies for economic and personal adaptation, emphasizing proactive approaches to remain valuable in the approaching AI-driven era.

Mostra di più

Mappa mentale

Mind Map

Domande frequenti

  • Who is Avital Balwit?

    Avital Balwit is the Chief of Staff at Anthropic, a leading AI lab.

  • What does Avital Balwit predict about the future of work?

    She predicts that AI will significantly impact employment, potentially leading to the end of traditional work as AI automates many tasks.

  • Why should we be concerned about AI advancements?

    AI advancements might lead to a future where many jobs are automated, creating challenges in employment and the need for new skills.

  • Which industries might be first affected by AI automation?

    Online and remote work industries, such as copywriting, tax preparation, and customer services, might be heavily automated first.

  • How might AI affect manual labor jobs compared to desk jobs?

    Manual labor jobs might take longer to be affected by AI due to the complexity of automating physical, hands-on tasks.

  • What industries are considered safer from AI automation?

    Regulated industries, such as medicine and law, are likely safer due to strict regulatory requirements and the human touch required.

  • How does unemployment affect people psychologically according to the video?

    Unemployment can lead to poorer health and increased stress, but shared experiences like a mass layoff might reduce individual distress.

  • What does the video say about societal needs beyond employment?

    Even with material abundance, people have an inherent need to contribute to society and find purpose beyond work.

  • What is the significance of Universal Basic Income (UBI) discussed in the video?

    UBI is considered as a potential solution to meet basic financial needs in a future with fewer employment opportunities.

  • What is a critical takeaway about preparing for the future of work?

    It's advisable to invest in skills and career paths less likely to be automated and stay informed about AI developments.

Visualizza altre sintesi video

Ottenete l'accesso immediato ai riassunti gratuiti dei video di YouTube grazie all'intelligenza artificiale!
Sottotitoli
en
Scorrimento automatico:
  • 00:00:00
    so this video is rather important
  • 00:00:01
    because it discusses the future of
  • 00:00:03
    employment and it's not just a random
  • 00:00:05
    person this is from avatal balwit and
  • 00:00:08
    this is someone who actually works at
  • 00:00:09
    the frontier lab and thropic and her
  • 00:00:12
    insights into how the future of work and
  • 00:00:15
    employment is going to change are very
  • 00:00:17
    important for individuals like yourself
  • 00:00:19
    and of course me so I'm going to do a
  • 00:00:20
    deep dive on this entire thing because I
  • 00:00:22
    truly believe that this is really
  • 00:00:24
    important that you can stay ahead of the
  • 00:00:25
    future now if you enjoy videos that are
  • 00:00:27
    specifically focusing on how you can use
  • 00:00:29
    AI to better your life and of course
  • 00:00:31
    preparing for AGI such as how to invest
  • 00:00:33
    in Ai and benefit from the trillions of
  • 00:00:35
    dollars that are flowing into that
  • 00:00:36
    industry don't forget to check out my
  • 00:00:38
    post AGI preparedness group this is
  • 00:00:39
    where videos like this actually get
  • 00:00:41
    uploaded a few days earlier but without
  • 00:00:43
    wasting any more time let's get straight
  • 00:00:44
    back to the video because it's really
  • 00:00:45
    important so essentially this is a post
  • 00:00:48
    about my last 5 years of work now if you
  • 00:00:51
    aren't understanding the significance of
  • 00:00:53
    this trust me at least watch the first
  • 00:00:55
    10 minutes because it's truly truly
  • 00:00:57
    incredible so essentially what we have
  • 00:01:00
    here ladies and gentlemen is a post by
  • 00:01:01
    Avital balwit and this is anthropics
  • 00:01:04
    Chief of Staff okay and this is kind of
  • 00:01:07
    like a personal reflection SL diary on
  • 00:01:10
    how the next few years of you know the
  • 00:01:12
    economy are going to go for the average
  • 00:01:14
    person in regards to work and where they
  • 00:01:16
    fit into the economy and how things are
  • 00:01:19
    going to be working now the reason I'm
  • 00:01:21
    covering this is because I think this is
  • 00:01:24
    probably one of the most remarkable
  • 00:01:26
    insights that we are lucky to have to
  • 00:01:28
    get because a lot of these you know
  • 00:01:30
    people at Frontier labs they truly don't
  • 00:01:33
    talk about how the future of work is
  • 00:01:34
    going to be so it is something that I
  • 00:01:36
    think is very very important for us to
  • 00:01:38
    discuss because this is something that
  • 00:01:41
    is truly truly pivotal especially for
  • 00:01:44
    this Monumental period of time so there
  • 00:01:46
    are four main things I'm going to Tim
  • 00:01:47
    stamp them in the below but just take a
  • 00:01:49
    look at this so she says I am 25 okay
  • 00:01:53
    and the these next three years might be
  • 00:01:55
    the last few years that I work okay
  • 00:01:57
    remember this is anthropics chief of
  • 00:01:59
    staff and she says that the next 3 years
  • 00:02:01
    might be the last few years that I
  • 00:02:03
    actually work which is number one an
  • 00:02:05
    incredible statement but number two it
  • 00:02:08
    should show you the gravity of what we
  • 00:02:10
    are truly dealing with here it says I am
  • 00:02:12
    not ill nor am I becoming a stay-at-home
  • 00:02:14
    mom nor have I been so financially
  • 00:02:16
    fortunate to be on the brink of
  • 00:02:17
    voluntary retirement I stand at the age
  • 00:02:20
    of a technological development that
  • 00:02:22
    seems likely should it arrive to end
  • 00:02:25
    employment as I know it so this is not
  • 00:02:28
    some kind of clickbait title where I'm
  • 00:02:30
    trying to draw you in this is the exact
  • 00:02:32
    words from this person who is working
  • 00:02:35
    actively at anthropic now anthropic is
  • 00:02:37
    one of the leading air Labs that made
  • 00:02:40
    Claude 3 which is of course Claude Opus
  • 00:02:42
    which is one of the frontier models that
  • 00:02:44
    is performing very effectively now what
  • 00:02:47
    we can see here is something absolutely
  • 00:02:49
    incredible we can see them stating
  • 00:02:51
    clearly that if it does arrive okay you
  • 00:02:54
    know where whe we're standing at the
  • 00:02:55
    edge of this which seems likely that
  • 00:02:58
    it's going to end employment so so you
  • 00:03:00
    know how a lot of the time okay
  • 00:03:02
    individuals in certain communities they
  • 00:03:04
    may discuss labor and work as something
  • 00:03:07
    that of course is you know going to be
  • 00:03:09
    there for many individuals like you and
  • 00:03:11
    myself but the prevailing trend from
  • 00:03:13
    many of the industry leaders that I
  • 00:03:15
    continue to see is the fact that there
  • 00:03:18
    won't be any true work after a certain
  • 00:03:20
    period of time due to the fact that we
  • 00:03:22
    have a situation on our hands where the
  • 00:03:25
    technological developments that are
  • 00:03:27
    coming are most likely to end this
  • 00:03:29
    Paradigm hence the reason post AGI
  • 00:03:32
    preparedness so that you can navigate
  • 00:03:33
    this now basically she then States I
  • 00:03:36
    work at a frontier AI company with every
  • 00:03:38
    iteration of our model I am confronted
  • 00:03:41
    with something more capable and more
  • 00:03:43
    General than before that's just
  • 00:03:45
    absolutely crazy with every iteration of
  • 00:03:47
    our model I'm confronted with something
  • 00:03:50
    more capable and more General before at
  • 00:03:52
    this stage it can completely generate
  • 00:03:54
    content on a wide range of topics it can
  • 00:03:56
    summarize and analyze text possibly well
  • 00:03:59
    and as someone who once made money at
  • 00:04:02
    one point in time as a freelance writer
  • 00:04:04
    and prided myself on my ability to write
  • 00:04:06
    large amounts of content quickly a skill
  • 00:04:08
    which like cutting blocks of ice from a
  • 00:04:10
    frozen ponds is arguably obsolete I find
  • 00:04:14
    it hard to not notice these advances
  • 00:04:16
    okay remember that one I find it hard to
  • 00:04:18
    not notice these advances freelance
  • 00:04:21
    writing was always an oversubscribed
  • 00:04:23
    skill set and the introduction of
  • 00:04:25
    language models has further intensified
  • 00:04:27
    the competition now remember the whole
  • 00:04:29
    point of the these videos is to ensure
  • 00:04:31
    that you are paying attention because a
  • 00:04:33
    lot of people just say ah you know we'll
  • 00:04:35
    find new jobs we'll be fine y y y y and
  • 00:04:37
    I think that's very very ignorant
  • 00:04:39
    because it's quite hard to ignore the
  • 00:04:42
    kind of developments that are going and
  • 00:04:44
    eventually they will be you know
  • 00:04:46
    impacting nearly every industry on the
  • 00:04:48
    planet so it says the general re
  • 00:04:50
    reaction okay and this is where you know
  • 00:04:52
    this is exactly what I say the general
  • 00:04:54
    reaction to language models among
  • 00:04:55
    knowledge workers is one of denial okay
  • 00:04:58
    one of denal
  • 00:05:00
    okay and this is why I'm making this
  • 00:05:01
    video because I want you all to
  • 00:05:03
    understand that whilst yes in previous
  • 00:05:06
    technological revolutions like when we
  • 00:05:08
    had the industrial revolution of the
  • 00:05:10
    farming industry and many people were
  • 00:05:13
    thinking oh no all of these computers
  • 00:05:15
    are going to take our jobs and all of us
  • 00:05:17
    that work in the farm industry come the
  • 00:05:19
    problem is is that we're of course going
  • 00:05:21
    to be having a situation where no longer
  • 00:05:23
    are we going to be needed in the farming
  • 00:05:25
    industry so we might lose our place in
  • 00:05:28
    society but of course in that stage we
  • 00:05:30
    did have a lot more things to do but the
  • 00:05:32
    reason that this time is different okay
  • 00:05:35
    and you shouldn't be in denial okay
  • 00:05:37
    especially among knowledge workers is
  • 00:05:39
    that AI is the automation of automation
  • 00:05:42
    so it's a feedback loop because AI is
  • 00:05:44
    something that can automate itself which
  • 00:05:46
    is different than previous generations
  • 00:05:50
    remember that this time is different so
  • 00:05:52
    it says they grasp at the ever
  • 00:05:53
    diminishing number of places where such
  • 00:05:55
    models still struggle rather than
  • 00:05:57
    noticing the ever growing range of tasks
  • 00:06:00
    where they still have reached or pass
  • 00:06:02
    human level this is something that I
  • 00:06:04
    truly truly think people need to
  • 00:06:06
    understand it's about how you view the
  • 00:06:08
    AI problem okay many people literally
  • 00:06:11
    just focus on the fact that llms you
  • 00:06:13
    know hallucinate once or twice or in
  • 00:06:15
    generative AI images sometimes they make
  • 00:06:17
    mistakes but you kind of look at things
  • 00:06:19
    and you're like wait overall okay like
  • 00:06:21
    if we take the entire pie of AI into
  • 00:06:24
    account overall there's like a small
  • 00:06:26
    subsection that with every iteration is
  • 00:06:28
    growing and growing and growing and
  • 00:06:30
    growing whereas like most people are
  • 00:06:32
    like oh no it can't do this small bit oh
  • 00:06:33
    it can still not do this small bit oh it
  • 00:06:35
    can still not do this small bit and
  • 00:06:36
    eventually what's going to happen is
  • 00:06:38
    that it's all going to be able to be
  • 00:06:39
    done by AI which means that like if
  • 00:06:41
    you're at this stage right now where you
  • 00:06:43
    can see that okay AI is slowly starting
  • 00:06:46
    to surpass human cognitive capabilities
  • 00:06:48
    it's best at this point to be like okay
  • 00:06:51
    I'm going to you know pay attention here
  • 00:06:53
    and at least position myself rather than
  • 00:06:54
    being as she states in a place of denial
  • 00:06:58
    so you know she says many will point out
  • 00:07:00
    that AI systems are not yet writing
  • 00:07:02
    award-winning books let alone patenting
  • 00:07:05
    inventions but and this is the most
  • 00:07:07
    important thing most of us also don't do
  • 00:07:10
    these things okay and this is what I'm
  • 00:07:11
    stating that a lot of times people like
  • 00:07:13
    oh AI can't write you know an entire
  • 00:07:15
    novel from scratch and AI can't you know
  • 00:07:17
    paint this and it can't make this poster
  • 00:07:19
    you know uh coherently with the text in
  • 00:07:21
    5 seconds like I could but you couldn't
  • 00:07:23
    do that most people couldn't do that AI
  • 00:07:25
    is able to do a lot more than the
  • 00:07:27
    average person which is the point okay
  • 00:07:29
    most people don't do these things and
  • 00:07:31
    this is something that most people are
  • 00:07:33
    failing to acknowledge okay the
  • 00:07:36
    economically and politically relevant
  • 00:07:38
    comparison on most TOS is not whether
  • 00:07:40
    the language model is better than the
  • 00:07:42
    best human it's whether they are better
  • 00:07:44
    than the human that would otherwise do
  • 00:07:47
    the task and this is the most important
  • 00:07:49
    distinction I'm going to say it again
  • 00:07:51
    the economically and politically
  • 00:07:53
    relevant comparison on most tasks is not
  • 00:07:56
    whether the language model is better
  • 00:07:58
    than the best human it's whether they
  • 00:08:00
    are better than the human that would
  • 00:08:02
    otherwise do the tasks basically this
  • 00:08:05
    question is the best question okay
  • 00:08:07
    because it shows you at what point
  • 00:08:09
    things get automated and this is what
  • 00:08:11
    I've spoken about before imagine this
  • 00:08:13
    for example right you have someone who
  • 00:08:15
    works in a store okay let's say that
  • 00:08:17
    person previously they did you know
  • 00:08:19
    checkout at a grocery store okay right
  • 00:08:21
    now we have ai self checkouts or
  • 00:08:23
    whatever we've got all these kind of
  • 00:08:24
    crazy things going on of course the self
  • 00:08:26
    checkout is just a piece of technology
  • 00:08:28
    but the point is guys is that is it
  • 00:08:31
    sufficient to have an AI system be
  • 00:08:34
    better than the average person that
  • 00:08:36
    would do the task in any scenario where
  • 00:08:38
    that is true you are going to get
  • 00:08:40
    automation of that labor it doesn't need
  • 00:08:42
    to be better than the best person it
  • 00:08:45
    just needs to be better than the average
  • 00:08:46
    task that it would take and that is the
  • 00:08:48
    important distinction because a lot of
  • 00:08:50
    people think oh this thing can't create
  • 00:08:52
    Hollywood blockbuster movies it's not
  • 00:08:53
    going to you know affect the movie
  • 00:08:55
    industry but it will affect other
  • 00:08:57
    Industries like the film industry the
  • 00:08:59
    photography industry all of these other
  • 00:09:01
    ones especially when you know we look at
  • 00:09:03
    Future models how much smarter they
  • 00:09:05
    going to be just because they're not
  • 00:09:07
    going to be as smart as Einstein it
  • 00:09:09
    doesn't mean they're going to be smarter
  • 00:09:10
    not smarter than the average researcher
  • 00:09:13
    which is of course very fascinating and
  • 00:09:15
    then it says this makes the objection
  • 00:09:17
    that AI systems are not yet coding long
  • 00:09:19
    sequences are doing more than fairly
  • 00:09:21
    basic math on their own a more relevant
  • 00:09:23
    one but these systems will continue to
  • 00:09:26
    improve at all cognitive tasks okay and
  • 00:09:29
    that's an important distinction because
  • 00:09:31
    a lot of the cognitive labor you know
  • 00:09:33
    it's kind of going away from the trends
  • 00:09:35
    that we can see okay the shared goal of
  • 00:09:37
    the field of artificial intelligence is
  • 00:09:39
    to create a system that can do anything
  • 00:09:42
    and she says here that I expect us to to
  • 00:09:44
    soon reach it and if I'm right how
  • 00:09:47
    should we think about the coming
  • 00:09:48
    adolescence of work so now here's where
  • 00:09:51
    we talk about the future of society so
  • 00:09:52
    she says that it's not worth noting that
  • 00:09:54
    upfront that even today work is far from
  • 00:09:56
    the only way to participate in society
  • 00:09:59
    but nevertheless it has proven to be the
  • 00:10:01
    best way to transfer wealth and
  • 00:10:02
    resources it provides personal Goods
  • 00:10:04
    like social connection status and
  • 00:10:06
    meaning and of course it offers social
  • 00:10:08
    goods like political stability and
  • 00:10:10
    here's where we get into the crazy
  • 00:10:11
    things given this should we meet the
  • 00:10:13
    possibility of its loss with sadness
  • 00:10:15
    fear and Joy or hope the overall
  • 00:10:17
    economic effects of AGI are difficult to
  • 00:10:20
    forecast and here I will focus on the
  • 00:10:22
    question of how people will feel without
  • 00:10:25
    work whether they will or can be happy
  • 00:10:27
    and there are obviously other vital
  • 00:10:29
    questions like how will people be able
  • 00:10:32
    to meet their material needs many have
  • 00:10:34
    examined this question and this is the
  • 00:10:36
    important part here okay and this is why
  • 00:10:37
    I launched a community and why we're
  • 00:10:39
    working on all these things many have
  • 00:10:41
    examined this question okay with no
  • 00:10:44
    final answer yet as adopted by official
  • 00:10:46
    policy for this contingency by any
  • 00:10:49
    government so there is not a single
  • 00:10:50
    government that is currently looking at
  • 00:10:52
    this problem and thinking okay this is
  • 00:10:54
    going to be an issue in a few years how
  • 00:10:55
    are we going to tackle this and I think
  • 00:10:57
    that that is important because many
  • 00:10:59
    people think that okay the government is
  • 00:11:01
    going to be there to take care of me but
  • 00:11:03
    I would say that the government aren't
  • 00:11:05
    really babysitters they just kind of
  • 00:11:06
    look over you every now and again to see
  • 00:11:08
    if you're doing all right and if you're
  • 00:11:09
    not about to die the government doesn't
  • 00:11:11
    really care about you that much of
  • 00:11:13
    course depending on where you live this
  • 00:11:15
    is something that can be argued of
  • 00:11:17
    course but I do think that this is for
  • 00:11:19
    the majority of societies and the
  • 00:11:21
    majority of things it is very very true
  • 00:11:23
    and of course you can see I will go
  • 00:11:25
    ahead and assume that people can meet
  • 00:11:27
    their needs through Universal basic
  • 00:11:28
    income or other transfers and will
  • 00:11:30
    solely concentrate on the question of
  • 00:11:32
    whether people can and will be happy or
  • 00:11:34
    at least as happy as they are now
  • 00:11:35
    without work so here's where we talk
  • 00:11:37
    about the obsolescence of knowledge work
  • 00:11:40
    so you can see I expect AI to get much
  • 00:11:43
    better than it is today and research on
  • 00:11:44
    AI systems has shown that they
  • 00:11:46
    predictably improve given better
  • 00:11:48
    algorithms more and better quality data
  • 00:11:51
    and of course more computational power
  • 00:11:53
    labs are in the process of further
  • 00:11:55
    scaling up their clusters the groupings
  • 00:11:57
    of computers that the algorithms run on
  • 00:11:59
    and machine learning is a young field
  • 00:12:01
    with an enormous amount of lwh hanging
  • 00:12:03
    fuit in terms of discoveries I meant to
  • 00:12:05
    say fruit there um and this means that
  • 00:12:07
    you know researchers can continuously
  • 00:12:08
    find improvements on the algorithms of
  • 00:12:10
    these AI systems so the point here and I
  • 00:12:12
    think this is one that you know it kind
  • 00:12:14
    of goes back to what Leopold Ashen
  • 00:12:15
    brener said is that when we take a look
  • 00:12:18
    okay at the fact that there is a lot of
  • 00:12:19
    lwh hanging fruit this kind of shifts
  • 00:12:22
    our opinion it's not a field that is you
  • 00:12:24
    know mature to the point where you know
  • 00:12:26
    we are you know at the edge of
  • 00:12:28
    everything that we know or at least what
  • 00:12:29
    we think we know and it's very unlikely
  • 00:12:31
    we're going to get a breakthrough for
  • 00:12:32
    another 20 to 50 years like in some
  • 00:12:35
    Fields this is a field that you know
  • 00:12:37
    when we look at you know the fact that
  • 00:12:39
    if we know that we scale up another 10
  • 00:12:41
    times we're about to get a whole lot lot
  • 00:12:43
    more we know that if we improve the data
  • 00:12:46
    quality and the amount of data we're
  • 00:12:47
    about to get a whole lot more
  • 00:12:48
    improvements we also know that there's a
  • 00:12:50
    lot of inefficient algorithms that we do
  • 00:12:52
    use and we know that there are literally
  • 00:12:55
    a million different ways to improve
  • 00:12:57
    these models in terms of the algorithm
  • 00:12:59
    efficiencies in terms of how they
  • 00:13:00
    respond so the point here is that there
  • 00:13:03
    is a lot of low hanging fruit in this
  • 00:13:05
    industry and we're only at the frontier
  • 00:13:08
    of what these models are going to be
  • 00:13:09
    capable of since we haven't like
  • 00:13:11
    exhausted all of those things if we had
  • 00:13:13
    like exhausted the compute capabilities
  • 00:13:16
    uh all the algorithmic discoveries and
  • 00:13:18
    we were like at the wits end and these
  • 00:13:19
    are the kind of systems that we had then
  • 00:13:22
    maybe potentially it could be argued
  • 00:13:25
    that we were near the end in terms of
  • 00:13:26
    the growth especially in terms of the
  • 00:13:28
    Curve maybe the growth would you know
  • 00:13:30
    start to slow off like that so that is
  • 00:13:32
    why I think this is really important
  • 00:13:34
    because there is a lot of lwh hanging
  • 00:13:36
    through that could truly give us a lot
  • 00:13:39
    more exponential gains and you can see
  • 00:13:41
    while there is an enormous amount of
  • 00:13:43
    data that has already been fed through
  • 00:13:45
    them there is still more to be found and
  • 00:13:47
    it can also be generated by the systems
  • 00:13:49
    themselves so given the scaling laws we
  • 00:13:52
    can reasonably foresee that these
  • 00:13:54
    systems will keep getting better at
  • 00:13:56
    least until these inputs run out so what
  • 00:13:59
    rate will they get better language
  • 00:14:01
    models are not for the most part
  • 00:14:02
    continuously improving they get better
  • 00:14:05
    in discontinuous jumps a rough analogy
  • 00:14:08
    to the current llm process is that
  • 00:14:10
    making a new model is like baking a cake
  • 00:14:12
    you figure out your data and algorithms
  • 00:14:14
    like mixing the batter then you
  • 00:14:16
    pre-train the model that is to run it on
  • 00:14:18
    a large number of computers for several
  • 00:14:20
    months like put it in the oven then you
  • 00:14:22
    do some posttraining like frosting and
  • 00:14:24
    decorating the cake and of course this
  • 00:14:25
    can adjust the model in certain
  • 00:14:27
    different ways to make it more hard less
  • 00:14:29
    honest or to make it good at certain
  • 00:14:31
    specific skills or use cases but what
  • 00:14:34
    matters to most of us for the model's
  • 00:14:36
    capabilities at least right now of
  • 00:14:38
    course is the underlying cake and this
  • 00:14:40
    can't be easily adjusted without
  • 00:14:41
    starting over and baking something
  • 00:14:43
    entirely new so when it comes to the
  • 00:14:45
    rate of progress when these models seem
  • 00:14:47
    to Plateau you should actually assume
  • 00:14:49
    that that just just means that the next
  • 00:14:51
    model is in the oven but just hasn't
  • 00:14:53
    come out yet and basically what they're
  • 00:14:55
    stating is that whil yes you know some
  • 00:14:57
    people do talk about how their
  • 00:14:59
    is this iterative deployment at open aai
  • 00:15:02
    from what we can see growth in llms is
  • 00:15:04
    kind of like this we hit a staircase we
  • 00:15:06
    hit a staircase and we hit a staircase
  • 00:15:08
    because essentially what you have is you
  • 00:15:09
    have let's say for example you have gbt
  • 00:15:11
    5 which brings us to this level then we
  • 00:15:12
    got gbt 4 then we get gbt 5 okay and the
  • 00:15:15
    reason it's kind of like that is because
  • 00:15:17
    like they said you have this middle bit
  • 00:15:19
    here where the model is being you know
  • 00:15:21
    trained whatever or you know mixed or
  • 00:15:23
    whatever you call it whatever and then
  • 00:15:24
    of course at the end you've got the
  • 00:15:25
    posttraining and then of course here
  • 00:15:27
    you've got the pre-training so this this
  • 00:15:29
    is essentially where you know you're
  • 00:15:30
    training the model for this entire
  • 00:15:32
    period of time so what's interesting is
  • 00:15:33
    that this period of time where you train
  • 00:15:35
    the model this entire period of time do
  • 00:15:37
    apologize for the colors there but
  • 00:15:38
    training the model and then of course
  • 00:15:40
    the model gets released after the
  • 00:15:41
    posttraining and then that's when we get
  • 00:15:43
    those new jumps which is why it's very
  • 00:15:45
    hard to say that right now you know
  • 00:15:47
    things are plateauing because these
  • 00:15:48
    models do take several months in most
  • 00:15:51
    cases to train especially because the
  • 00:15:53
    Clusters can't support that much uh you
  • 00:15:56
    know work going on at the moment so here
  • 00:15:58
    we can see many expect AI to eventually
  • 00:16:00
    be able to do every economically useful
  • 00:16:03
    task which is a truly insane statement
  • 00:16:05
    so many expect AI to do every
  • 00:16:07
    economically useful task which is
  • 00:16:09
    absolutely crazy and remember she agrees
  • 00:16:11
    now this is anthropics Chief of Staff
  • 00:16:13
    here and it says given the current
  • 00:16:16
    trajectory of technology I expect AI to
  • 00:16:18
    First excel at any kind of online work
  • 00:16:21
    essentially anything that a remote
  • 00:16:22
    worker can do AI will do better
  • 00:16:25
    copywriting tax preparation customer
  • 00:16:27
    services and many other tasks are or
  • 00:16:31
    will soon be heavily automated I can see
  • 00:16:33
    the beginnings in areas like software
  • 00:16:36
    development and contract law generally
  • 00:16:39
    tasks that involved reading analyzing
  • 00:16:41
    and synthesizing information and then
  • 00:16:43
    generating content based on it it seems
  • 00:16:45
    ripe for replacement by language models
  • 00:16:48
    so basically right here this gives you a
  • 00:16:50
    first insight into some of the careers
  • 00:16:52
    that are under not scrutiny but under
  • 00:16:54
    the microscope in terms of where things
  • 00:16:57
    are going to be truly automated you
  • 00:16:59
    might want to note this down I will have
  • 00:17:01
    of course all the notes on in my school
  • 00:17:03
    community so that you guys can easily
  • 00:17:04
    read this and digest the information and
  • 00:17:06
    an easy to use mattera but guys this is
  • 00:17:10
    some key key statements that people are
  • 00:17:12
    not paying attention to because the main
  • 00:17:14
    thing here is that anything on a
  • 00:17:15
    computer anything remote anything online
  • 00:17:18
    those things you have to think now about
  • 00:17:20
    okay if I'm a remote worker if I'm
  • 00:17:22
    someone that's enjoying my life
  • 00:17:23
    traveling around the world or that's
  • 00:17:24
    what I want to do I need to think about
  • 00:17:26
    these industries here and how I can
  • 00:17:29
    position myself okay to not be automated
  • 00:17:31
    away or to not at least be less
  • 00:17:34
    economically valuable than an AI system
  • 00:17:37
    sure you might be 10% better than an AI
  • 00:17:39
    system but you have to remember that
  • 00:17:41
    it's much easier for a company to
  • 00:17:43
    replace your high salary with an AI
  • 00:17:45
    system that costs much less and I do
  • 00:17:48
    think in the future what we will see is
  • 00:17:51
    AI systems that are truly truly
  • 00:17:53
    effective and of course that can run for
  • 00:17:55
    24 hours remember you're just a human
  • 00:17:58
    you can't run for 24 hours like these AI
  • 00:18:00
    systems can and with all the AI agents
  • 00:18:02
    being built on the back end trust me
  • 00:18:04
    guys things are truly right for
  • 00:18:06
    disruption so you need to think okay how
  • 00:18:08
    can I do this okay and how can I manage
  • 00:18:11
    myself better in order to at least not
  • 00:18:13
    be in that place where I'm less
  • 00:18:15
    economically valuable than this thing
  • 00:18:17
    and here's where we have some more stuff
  • 00:18:19
    so here's where we get to the
  • 00:18:20
    obsolescence part and this is where we
  • 00:18:22
    actually start to look at certain
  • 00:18:23
    industries and how things are going to
  • 00:18:26
    be changing so essentially what she told
  • 00:18:28
    about here is that you know this is not
  • 00:18:30
    going to come for every industry at the
  • 00:18:32
    same Pace it's not like one day
  • 00:18:34
    everyone's working and the next day
  • 00:18:36
    everyone's not we're going to have a
  • 00:18:37
    situation on our hands where this area
  • 00:18:39
    is now affected this area is now
  • 00:18:41
    affected this area is now affected and
  • 00:18:43
    it's going to definitely be proportional
  • 00:18:45
    to AI system releases for example when
  • 00:18:47
    GPT 5 does release I do expect a wider
  • 00:18:50
    range of jobs to be not completely
  • 00:18:52
    automated but very nearly on that
  • 00:18:55
    Chopping Block Level to where you know
  • 00:18:57
    there's going to be another wave of
  • 00:18:58
    potentially layoffs or potentially you
  • 00:19:00
    know companies that used to hire people
  • 00:19:02
    not hiring people and the problem is is
  • 00:19:04
    that these effects are Downstream
  • 00:19:06
    meaning that if someone was going to
  • 00:19:08
    create a company in the future and they
  • 00:19:10
    choose not to create a company you don't
  • 00:19:13
    experience that level until like maybe
  • 00:19:15
    12 to 48 months after because the
  • 00:19:17
    companies that would have replaced the
  • 00:19:19
    old ones that wouldn't hire you those
  • 00:19:21
    ones no longer exist so overall that
  • 00:19:23
    entire market share has died down so
  • 00:19:25
    essentially um the pace of improvement
  • 00:19:28
    in robotics lag significantly behind
  • 00:19:30
    cognitive automation it's improving as
  • 00:19:32
    well but more slowly so this is
  • 00:19:34
    basically what they're saying is that if
  • 00:19:35
    you do anything with your hands that is
  • 00:19:37
    pretty intricate um in a guided specific
  • 00:19:40
    situation for example you know surgeons
  • 00:19:43
    gardeners plumbers jewelry makers hair
  • 00:19:46
    stylists as well as those who repair
  • 00:19:48
    iron work or who make stained glass
  • 00:19:50
    might find their handiwork contributing
  • 00:19:52
    to our society for many more years to
  • 00:19:54
    come and this is of course uh you know
  • 00:19:57
    very very true I do think that you know
  • 00:19:59
    people in these industries they have a
  • 00:20:01
    unique set of skills that you know for a
  • 00:20:03
    robot to do as well as a plumber it's
  • 00:20:05
    very very hard considering you know the
  • 00:20:07
    fact that it needs to navigate its way
  • 00:20:09
    into a house deal with the customer be
  • 00:20:11
    able to identify what's going wrong stop
  • 00:20:13
    the water flowing uh whatever issue it
  • 00:20:15
    may be then coordinate with all of these
  • 00:20:17
    kind of things like if you try to get a
  • 00:20:20
    team of you know Engineers to to get a
  • 00:20:22
    robot to do that that would cost a lot
  • 00:20:24
    and it would be so expensive to do that
  • 00:20:26
    but you know an AI system that's on a
  • 00:20:28
    computer running that can take control
  • 00:20:30
    of a computer that is something that is
  • 00:20:32
    vastly easier which is why a lot of the
  • 00:20:34
    people who work at desks are more at
  • 00:20:36
    risk than someone who does a manual
  • 00:20:39
    labor job and this is something that I
  • 00:20:41
    think this trend is important to pay
  • 00:20:43
    attention to so of course regulated
  • 00:20:45
    Industries like medicine or the Civil
  • 00:20:47
    Service will have human involvement per
  • 00:20:48
    longer but even there I expecting
  • 00:20:50
    increasingly number of human workers who
  • 00:20:53
    are you know supplemented with AI
  • 00:20:55
    systems working alongside them so
  • 00:20:57
    basically with this one is that if you
  • 00:20:58
    work in a regulated industry or you're
  • 00:21:00
    thinking about getting into one I think
  • 00:21:02
    those ones are still pretty good for the
  • 00:21:03
    future because regulatory boards and
  • 00:21:06
    stuff like that they do take a lot of
  • 00:21:07
    time to allow people to get into their
  • 00:21:10
    Industries you know for example like to
  • 00:21:11
    become a doctor there's all these tests
  • 00:21:13
    you have to take to become a lawyer you
  • 00:21:15
    got to pass the bar you got to do this
  • 00:21:17
    it's this whole kind of thing whether
  • 00:21:19
    you want to call it status even if the
  • 00:21:21
    even if the AI system is better there's
  • 00:21:23
    just this human kind of aspect where
  • 00:21:26
    even if AI systems are better because of
  • 00:21:28
    the regulatory bodies and how AI might
  • 00:21:30
    be viewed at that time AI is probably
  • 00:21:32
    not going to be penetrating those
  • 00:21:34
    Industries very very soon like it's not
  • 00:21:36
    going to be okay AI here boom let's use
  • 00:21:38
    it there's still regulatory bodies that
  • 00:21:40
    people you know follow and it's going to
  • 00:21:42
    be really hard to change those because
  • 00:21:44
    small changes and small issues with AI
  • 00:21:46
    systems unless they're like 100% um and
  • 00:21:49
    unless they have no bias no this no that
  • 00:21:51
    no that um it's going to be really hard
  • 00:21:53
    to change those Legacy Industries
  • 00:21:55
    because they are very very resistant to
  • 00:21:58
    change and they were built upon those
  • 00:22:00
    you know strict foundations so you know
  • 00:22:03
    careers that have like you know all
  • 00:22:04
    those regulations around them I think
  • 00:22:05
    those are going to be secure for a very
  • 00:22:07
    very longer time now of course we talk
  • 00:22:09
    about industries that are going to be
  • 00:22:10
    preferred for humans so for example you
  • 00:22:13
    know the those jobs that you know where
  • 00:22:15
    you want a human to do it even if an AI
  • 00:22:18
    system might be better so for example
  • 00:22:20
    jobs that might fall into this category
  • 00:22:22
    include counselors duelers caretakers
  • 00:22:24
    for the elderly babysitters Preschool
  • 00:22:27
    teachers priests and religious leaders
  • 00:22:30
    even sex workers much has been made of
  • 00:22:32
    AI girlfriends but I still expect that a
  • 00:22:34
    large percentage of buyers of impers and
  • 00:22:36
    sexual services will have a strong
  • 00:22:38
    preference for humans and I think this
  • 00:22:40
    is an interesting term here nostalgic
  • 00:22:42
    jobs and I think that term actually does
  • 00:22:44
    make sense so you can see that these
  • 00:22:46
    jobs right here even if you have like an
  • 00:22:48
    AI system that is you know just amazing
  • 00:22:51
    you know sitting a baby like let's say
  • 00:22:53
    it's great as a babysitter I would not
  • 00:22:55
    want an AI system babysitting my child I
  • 00:22:57
    would want it to have the human- to
  • 00:22:59
    human connection I would want it to
  • 00:23:01
    understand what other humans want this
  • 00:23:02
    is just going to be something that maybe
  • 00:23:04
    as I grow older and as I you know get
  • 00:23:06
    older into my older years I'm just like
  • 00:23:08
    that old person who's who's scared of
  • 00:23:10
    robotic technology and I'm like no I
  • 00:23:12
    want a human looking after my kid or
  • 00:23:14
    whatever or maybe you know even people
  • 00:23:15
    growing up now they still find it weird
  • 00:23:17
    but over time those values can shift as
  • 00:23:21
    robots become more and more useful and
  • 00:23:23
    as they become more and more humanlike
  • 00:23:25
    so it's going to be interesting to see
  • 00:23:27
    how things are going to be you know done
  • 00:23:29
    in the aspect but these jobs right here
  • 00:23:31
    they're essentially valued because there
  • 00:23:33
    is a human right there so if I was going
  • 00:23:34
    through a problem I wouldn't really want
  • 00:23:36
    to talk to an AI system because the
  • 00:23:38
    whole thing about like counseling and
  • 00:23:40
    stuff like that and being you know
  • 00:23:41
    talking to a therapist is that there's a
  • 00:23:43
    human who actually empathizes with you
  • 00:23:45
    and no matter how smart an AI system
  • 00:23:47
    might be it hasn't had its own you know
  • 00:23:50
    subjective or objective whatever you
  • 00:23:52
    want to say experience of the real world
  • 00:23:54
    so it's not going to be able to
  • 00:23:55
    empathize with you on that way that an
  • 00:23:57
    actual person that can be like oh you
  • 00:23:58
    know I I understand that I went through
  • 00:24:00
    that you know I've had this experience I
  • 00:24:02
    had that happen to me when I was 21 um
  • 00:24:03
    and that like things like that are just
  • 00:24:05
    completely Priceless so those things
  • 00:24:07
    there like those areas and categories
  • 00:24:09
    are going to really are going to really
  • 00:24:11
    do well which is why it's important to
  • 00:24:13
    like kind of look at this because you
  • 00:24:14
    know a lot of people don't understand
  • 00:24:16
    truly the dynamic on how things are
  • 00:24:18
    going to change given the nature of AI
  • 00:24:21
    going now for the rest of this article
  • 00:24:23
    there are some you know interesting
  • 00:24:25
    things there is the psychology of
  • 00:24:27
    employment which is you know as
  • 00:24:28
    automation rolls out across these
  • 00:24:30
    industries how are people going to feel
  • 00:24:32
    and basically what they're stating here
  • 00:24:34
    is that you know evidence shows
  • 00:24:36
    unsurprisingly okay in fact no this is
  • 00:24:38
    actually quite surprising is that you
  • 00:24:40
    know basically they're trying to look at
  • 00:24:41
    how unemployment actually affects people
  • 00:24:44
    okay and it says that one study that
  • 00:24:45
    tried to tackle this looked at the
  • 00:24:47
    effects of unemployment caused by the
  • 00:24:48
    collapse of the Spanish construction
  • 00:24:50
    industry on mental and physical health
  • 00:24:52
    and this particularly study was
  • 00:24:54
    attempting to disentangle the causality
  • 00:24:56
    because people who lose their job during
  • 00:24:58
    a nationwide collapse of an industry
  • 00:25:00
    will avoid this election effect these
  • 00:25:02
    individuals are no more likely to have
  • 00:25:04
    mental mental or physical issues than
  • 00:25:06
    other members of the population and it
  • 00:25:08
    says by looking at Large Scale survey
  • 00:25:10
    responses before and after the crisis
  • 00:25:12
    they found that unemployment did appear
  • 00:25:14
    to increase the likelihood of reporting
  • 00:25:16
    poorer health and it says here it does
  • 00:25:18
    seem that overall unemployment makes
  • 00:25:21
    people sadder sicker and more anxious
  • 00:25:23
    but it isn't clear if this is an
  • 00:25:25
    inherent fact of unemployment or a
  • 00:25:27
    contingent want okay because basically
  • 00:25:29
    the problem is this right if you right
  • 00:25:31
    now you're unemployed okay why are you s
  • 00:25:34
    think about that for a second okay why
  • 00:25:35
    are you s you're s because number one
  • 00:25:37
    you might have less status because you
  • 00:25:39
    don't have a job and number two you know
  • 00:25:40
    you don't have any money okay you don't
  • 00:25:42
    have any money coming in so you might
  • 00:25:43
    think okay I'm not bringing any value
  • 00:25:44
    into society and I'm not making a living
  • 00:25:46
    for myself and of course I might be
  • 00:25:48
    homeless and yada yada yada and you can
  • 00:25:50
    see right here you've got all these
  • 00:25:52
    things that they talk about okay you
  • 00:25:53
    know the financial effects are probably
  • 00:25:55
    the main one like if I gave you right
  • 00:25:57
    now $10 million and I said you can never
  • 00:25:59
    work a day in your life you know you're
  • 00:26:00
    never allowed to have a job even if you
  • 00:26:02
    wanted one you just are not allowed to
  • 00:26:03
    work most people would laugh run off
  • 00:26:05
    with a 10 million and be like this guy
  • 00:26:07
    doesn't know what he's talking about and
  • 00:26:08
    that's the main point here is that it's
  • 00:26:11
    hard to understand whether or not humans
  • 00:26:13
    derive meaning from employment or it's
  • 00:26:16
    just the contingent effects okay and you
  • 00:26:18
    can see right here is that of course
  • 00:26:20
    this might not occur in the context of
  • 00:26:22
    universal basic income and of course
  • 00:26:24
    it's compounded with the shame aspect of
  • 00:26:26
    being fired or laid off when you know
  • 00:26:28
    you really feel like you should be
  • 00:26:29
    working as opposed to the context where
  • 00:26:31
    essentially all workers have been
  • 00:26:32
    displaced intuitively it seems that
  • 00:26:35
    there should be more negative
  • 00:26:37
    psychological effects from losing a job
  • 00:26:39
    in a way that feels like a personal
  • 00:26:41
    failing or that sets one apart from
  • 00:26:43
    one's peers versus losing a job in a
  • 00:26:45
    blameless way or at the same time and
  • 00:26:47
    the same manner as one's peers basically
  • 00:26:50
    what they're stating here is as well is
  • 00:26:51
    that whilst you might think that a lot
  • 00:26:53
    of people who lose their jobs during
  • 00:26:54
    this revolution might be sad if everyone
  • 00:26:57
    loses it at the at the same time like
  • 00:26:59
    people did in the pandemic people
  • 00:27:01
    actually don't increase their
  • 00:27:02
    psychological distress because
  • 00:27:03
    everyone's in the same boat you can see
  • 00:27:05
    here they found that individuals who
  • 00:27:07
    were temporary laid off in April of 2020
  • 00:27:09
    reported lower levels of distress
  • 00:27:11
    compared to their peers who remained
  • 00:27:13
    employed and you can see here the
  • 00:27:14
    widespread nature of layoff normalized
  • 00:27:16
    The Experience reducing the personal
  • 00:27:18
    blame and fostering a sense of shared
  • 00:27:20
    experience and of course Financial
  • 00:27:21
    strain was mitigated by government
  • 00:27:23
    support personal savings and reduced
  • 00:27:25
    spendings which buffed against potential
  • 00:27:27
    distress so basically the stating that
  • 00:27:28
    there might not be as much angst and
  • 00:27:31
    anxiety in the future if everyone is in
  • 00:27:33
    the same boat but I do think that this
  • 00:27:35
    is going to be Nuance because there are
  • 00:27:36
    going to be a lot of industries that are
  • 00:27:38
    affected before other ones as we
  • 00:27:40
    previously discussed knowledge work
  • 00:27:42
    cognitive labor jobs jobs on on our
  • 00:27:44
    computer those ones are going to be
  • 00:27:45
    automated heavily away first and it's
  • 00:27:47
    not going to be all at once some
  • 00:27:49
    companies are going to choose to
  • 00:27:50
    implement them some companies are going
  • 00:27:51
    to say nope we're 100% human run we're
  • 00:27:53
    never going to use AI that's going to be
  • 00:27:55
    their marketing gimmick those are the
  • 00:27:56
    companies that you kind of want to you
  • 00:27:58
    know work for and be you know you know
  • 00:27:59
    aligning your options with but it's
  • 00:28:01
    going to be kind of interesting as well
  • 00:28:03
    like you know if you're looking to work
  • 00:28:04
    for a company you know try and see what
  • 00:28:05
    their CEO's view are on AI of course
  • 00:28:07
    every CEO's incentive is to make more
  • 00:28:09
    money so they're probably going to do
  • 00:28:11
    that but you'd be surprised at how
  • 00:28:12
    things work in the future and this is
  • 00:28:14
    where we get into two of the most
  • 00:28:15
    interesting Concepts here okay um and
  • 00:28:17
    this is where she talks about inbank the
  • 00:28:19
    culture books and this is where you know
  • 00:28:21
    they are completely post scarcity
  • 00:28:23
    Society money is viewed as crude and
  • 00:28:26
    Irrelevant for allocating resources
  • 00:28:28
    resources living space raw materials and
  • 00:28:30
    energy are produced in abundance for its
  • 00:28:32
    citizens and the capacity of its means
  • 00:28:34
    of production are ubiquitously and
  • 00:28:36
    comprehensively exceeded every
  • 00:28:38
    reasonable demands it's not
  • 00:28:40
    unimaginative citizens could make yet
  • 00:28:42
    culture at least has one need that this
  • 00:28:45
    abundance cannot satisfy that feeling
  • 00:28:47
    was the urge to not feel useless
  • 00:28:49
    basically to cut the nonsense this
  • 00:28:50
    passage highlights the fundamental and
  • 00:28:52
    AI need to feel like they are
  • 00:28:54
    contributing to something larger than
  • 00:28:56
    themselves even in a paradise where all
  • 00:28:59
    material needs are met the psychological
  • 00:29:01
    need still persists and she states here
  • 00:29:05
    that you know you can think about it
  • 00:29:06
    right now do you do anything that you
  • 00:29:08
    are notably worse at than other people
  • 00:29:10
    just for the sheer value of doing it she
  • 00:29:12
    talks about how you know being a
  • 00:29:13
    ballerina even though she's in her
  • 00:29:15
    mid-20s and you know being a ballerina
  • 00:29:17
    is long behind her but moving her body
  • 00:29:19
    like that still brings her joy and this
  • 00:29:21
    is where she talks about post AGI so she
  • 00:29:23
    says a renowned AI researcher once told
  • 00:29:26
    me that he is practicing for post AGI by
  • 00:29:28
    taking up activities that he's not
  • 00:29:30
    particularly good at Jiu-Jitsu surfing
  • 00:29:32
    and so on and savoring the doing even
  • 00:29:35
    without Excellence this is how we can
  • 00:29:37
    prepare for a future where we will have
  • 00:29:38
    to do things from Joy rather than need
  • 00:29:41
    where we will no longer be the best at
  • 00:29:43
    them but we will still have to choose
  • 00:29:44
    how we fill our days we will also not
  • 00:29:47
    need to choose how to fill our time
  • 00:29:48
    alone in the context where we are all
  • 00:29:50
    out of work and where this is one of our
  • 00:29:52
    main worries it means we built
  • 00:29:54
    relatively aligned artificial general
  • 00:29:57
    intelligence for the the same reasons I
  • 00:29:59
    expect us to reach AGI I expect it to
  • 00:30:01
    progress Beyond this point to where we
  • 00:30:03
    have superhuman systems for the same
  • 00:30:06
    reason these systems will be helpful of
  • 00:30:07
    anything we should expect that these
  • 00:30:09
    systems will be able to help with the
  • 00:30:11
    problems that they create and basically
  • 00:30:12
    what they're stating here is that if we
  • 00:30:14
    get superhuman systems that replace us
  • 00:30:16
    there's no reason to think that they
  • 00:30:18
    cannot help us if we don't have meaning
  • 00:30:20
    in this future world and I think this
  • 00:30:22
    article is very important like I said
  • 00:30:24
    I'm going to have like a you know a
  • 00:30:25
    short private blog where I summarize
  • 00:30:27
    everything into a nice on my school and
  • 00:30:29
    this video you know is likely going to
  • 00:30:30
    be released on the school Community
  • 00:30:32
    going to be released on the school
  • 00:30:33
    Community a few days earlier because
  • 00:30:35
    this is the community where we discuss
  • 00:30:37
    how we can actually navigate this stuff
  • 00:30:39
    and I don't want you guys to feel like
  • 00:30:40
    look there's nothing I can do there is
  • 00:30:42
    so much that you can do in order to be
  • 00:30:43
    proactive especially in the last few
  • 00:30:45
    years of work there are investments that
  • 00:30:47
    you can make there are ways that you can
  • 00:30:49
    transition your career from one area to
  • 00:30:51
    another there are mindsets that you can
  • 00:30:53
    use there are Frameworks that you can
  • 00:30:54
    apply to your life to ensure you're not
  • 00:30:56
    one of the people that get first
  • 00:30:58
    automated because whilst automation is
  • 00:31:00
    something that is coming it might not
  • 00:31:02
    get you first if you're able to move out
  • 00:31:04
    of the way and you're still able to be
  • 00:31:06
    relatively valuable in the economy
  • 00:31:08
    that's what the entire thing is about so
  • 00:31:10
    let me know what you think about this
  • 00:31:11
    entire document it is definitely a long
  • 00:31:13
    one maybe I did go over my time limit
  • 00:31:15
    here but I think this is probably one of
  • 00:31:16
    the most important articles because I
  • 00:31:18
    think by at least 2027 we're truly going
  • 00:31:21
    to see the trajectory of where things
  • 00:31:22
    are going because at that stage we will
  • 00:31:25
    likely have had at least four to 5 years
  • 00:31:27
    by 2030 of AI development we will likely
  • 00:31:30
    have the biggest clusters being built
  • 00:31:32
    you will likely have maxed out the you
  • 00:31:33
    know physical Hardware so we're likely
  • 00:31:36
    going to see where those developments
  • 00:31:38
    are going to be in the sense that any
  • 00:31:40
    more improvements are going to be hard
  • 00:31:41
    to come by and therefore we know that
  • 00:31:43
    the jobs that currently exist for humans
  • 00:31:45
    are probably going to be there for quite
  • 00:31:47
    some time so this was something that
  • 00:31:48
    most people just did Miss but I'm pretty
  • 00:31:50
    sure that this article was something
  • 00:31:52
    that I think is definitely valuable to
  • 00:31:53
    those of you who are concerned about the
  • 00:31:55
    future of post AI economics let me know
  • 00:31:57
    some of your thoughts on how all of this
  • 00:31:59
    is going to go down and what your plans
  • 00:32:01
    are and if you enjoy this video I'll see
  • 00:32:03
    you in the next
Tag
  • AI
  • Future of Work
  • Employment
  • Automation
  • Universal Basic Income
  • Psychological Impact
  • Skill Adaptation
  • Industry Impact