Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

AI is Coming ... #733

Open
nelsonic opened this issue Aug 30, 2019 · 81 comments
Open

AI is Coming ... #733

nelsonic opened this issue Aug 30, 2019 · 81 comments
Labels
discuss Share your constructive thoughts on how to make progress with this issue in-progress An issue or pull request that is being worked on by the assigned person starter A beginner-friendly issue that is a good starting point for a new person

Comments

@nelsonic
Copy link
Owner

nelsonic commented Aug 30, 2019

This is a post I'm developing ... It's been in my head for the last few years.
I'm going to need a few hours of focus time to flesh it out.
I really don't like writing "essays" in GitHub issues because the interface is not conducive to it.
I want to create my own outline editor where I can preview in real-time and expand/contract sections in both the writing and preview so that I can focus on specific section at a time ...

Ultimately my conclusion is this: you/we can either:
a) ignore AI and go about your life oblivious to fact that AI will fundamentally and irrevocably change the world in the next 5-10 years.
b) proactively learn to leverage AI and prepare for a future where Great Depression levels of unemployment lead to significant societal strife.


The future promised in science fiction is coming much sooner than most people realise or expect.

Most people do not actively think about the future for whatever reason. But everyone should! The future dominated by AI and automation will affect us all for better or worse.

image

The example of AI the media tend to focus on is self-driving cars or autonomous vehicles.
This is perhaps because Tesla is doing a good job of marketing their "autopilot" and more recently their "full self-driving" capabilities.

self-driving-object-detection

While we are unlikely to see see fully autonomous humanoid robots with general purpose AI such as The Gunslinger of Westworld (1973), Robocop (1987) or David of Spielberg's Ai (2001) in the next 5 years, we are going to see special purpose artificial intelligence accelerate significantly.

robocop-without-helmet

sofia-the-robot
Even though thousands of people/companies with billions in capital are working on human-like robots to make the technology more friendly and social for use in hospitality or nursing, e.g: the Sofia Robot https://en.wikipedia.org/wiki/Sophia_(robot) unveiled in 2016, this is actually a distraction because the real impact AI is going to have is not in the form of humanoid robots.

We already rely on automated machines to perform many jobs in a wide range of industries.

robot-arm-picking-tomatoes

Everything from food processing/packaging to car manufacturing. Assembly lines that once required thousands of people are now operated by hundreds of robots with minimal human supervision or intervention. The trend toward using machines to automate repetitive tasks is only going to accelerate and the jobs once performed by humans will done faster and cheaper by AI.

vehicle-manufacturing-robot

The Da Vinci surgical robot is assisting doctors to perform high precision surgery with minimal incision. The operations are all recorded and the data is analysed along with the outcome (success/failure) of the surgery. The footage is fed into a machine learning algorithm which learns how to perform surgery. This is great news for patients because the outcomes and post-surgical recover are improving because ops are less invasive. But the flip-side of this advancement is that fewer people are needed to perform the surgery. The surgical team of 6+ people
(surgeon, surgical assistant, surgical technologist, nurses, radiologist and anesthesiologist) is reduced to 3 people. Radiologist are rapidly being replaced by algorithms which are faster and more accurate. Anesthesia is calculated and applied by an algorithm. Surgical tech is a robot.

medical-surgery-robots

Most existing robots are single/special purpose; they do one thing well. Robots are faster, cheaper and more precise than humans and they don't get tired or sick.

How you feel about autonomous and respond to it is very much determined by your overall mindset. If you already feel like you are unable to learn new skills/technologies and adapt to change, you aren't going to enjoy the next 5-10 years ...

image

Consider the following two perspectives: "Artificial Intelligence (AI) is Coming and is going to":

1. Take Our Jobs and Make Us Obsolete

darryl-southpark-they-took-our-jobs

As Darryl in Southpark repeatedly states "they took our jobs"1. Estimates range between 25% and 40% of jobs are at risk of being automated by AI. It won't just be "coal mining and truck driving" as in Southpark, rather AI is going to "take" the jobs of everyone from retail workers to brain surgeons.

2. Free Us From Boring/Meaningless Jobs

rosie-jetsons-does-chores-for-family

Rosie the Robot in the Jetsons performed all the household chores freeing up the time of her human masters. If all the boring menial jobs like cleaning and food preparation are automated, that's great for the people who own the automated machines because they can get on with their lives.

But it's not going to be so great if the job you rely on for your income, which you use to pay rent/bills is automated and you are left unemployed and due to a lack of skills, unemployable.

Unemployment

The obvious menial tasks like retail checkout clerks, shelf-stackers and fast-food workers will be automated by robots in the next 5-10 years. But the not so obvious and traditionally well-paying jobs are also going to be automated by software and machine learning algorithms.

image
Robots And AI: The Future Is Automated And Every Job Is At Risk https://youtu.be/rnBAdnNIIXk

Traditional "white collar" jobs like Doctors, Bankers, Lawyers, Accountants are all being automated to varying degrees. Thousands of startups are actively working on automating the functions of:

  • Human Resources (HR) - the work that used to be done by a big team of people is now done by software and is increasingly being automated.
  • Administrative or Clerical jobs such as bank "back office" or loan officers.
  • Insurance underwriters - the people who input data, calculate premiums and pay out claims.
  • Legal clerks - why pay a human to read case files and precedents when an algorithm can read thousands of pages per minute and summarise the probability of a case being won cheaper and more effectively?
  • Tax Accountants - if all financial records are controlled by a system that automatically submits accurate tax returns on-time, what use is an accountant?

Comprehensive list: #318 (comment)

Self-Checkout > Amazon Go

The self-checkout that allow people to pay for purchases at grocery stores are a "stepping-stone" technology. They are clunky and (user) error prone.

image

Amazon Go uses AI to "scan" the person as they walk in and tracks people as they walk around the store adding items to their basket. People don't have to scan individual items at checkout, they simply transfer their desired purchases from their shopping basked/cart directly into carriers and payment is automatic.
image
https://www.theverge.com/2018/10/23/18010022/amazon-go-cashier-less-store-san-francisco-location-opens
Combined with automated shelf-stacking, a store visit can involve zero human interaction.

Google Assistant ("Duplex")

Google Duplex is an excellent example of how AI is already being used. Watch the I/O keynote from 2018 to see how call center work will be automated and millions of human telephone operators replaced by software.

google-duplex-make-real-phone-calls
https://youtu.be/ijwHj2HaOT0

If human interaction can be automated, you had better believe your job is not "safe".

More recently Google Duplex has "learned" how to use a car rental website through machine learning and can perform the entire booking based a single request to google assistant.
google-duplex-learns-the-internet
https://youtu.be/JbkoQGMf5DI

When the jobs people once did are automated, the unemployment line will grow ...

unemployment-line

will-work-for-food

Homelessness

People who do not have an income or savings or family they can stay with will be made homeless.

homeless-girl

homeless-mother-with-two-children

homeless-city-street

Loss of Meaning

Many people attach substantial significance and social status to their employment.
Some people even define themselves by what they do for a living. This is unsurprising since it's their primary activity in life and they have often invested much of their adult lives in it.
When people lose their jobs to AI, they will lose both social status and meaning.
In most cases people will have to settle for whatever job they can get in order pay their bills,
and almost invariably this will be a lower-paid and lower status job.

Riots

Economically displaced people may be incited to violence.

london-riots
https://en.wikipedia.org/wiki/2011_England_riots

We have seen this recently in Paris with the "yellow vest" protests:

paris-riots
https://www.voanews.com/europe/france-seeks-answers-after-police-failed-stop-paris-riots

Who Will AI Affect?

It can be difficult to conceptualise the loss of the job you are currently doing.
What you do might be a "niche" skill that you think nobody wants to automate.
But you're missing the point entirely. What happens to the billions of people who are currently doing other jobs when they are made redundant? Well, some of them might re-train to do your job. And since they have zero income, they might do your job for a lot less than you are currently being paid. So unless your are doing something that only you can do (e.g: your face is your "money maker" or "brand"), you should at least consider the possibility that you too will be replaced. And if you aren't worried about your job because you have concluded that you're irreplaceable, think about how society as a whole will change when there are many times more people out-of-work ...

Proxy-question: How many People did TV, the Internet or Smartphones Affect?

Perhaps a better question is: how many people did the invention/proliferation of Television affect?

TV?

65 Years ago only the wealthy elite could afford a television. In 1954 when the first color TV went on sale at $1,000, the equivalent of $8,500 in today's dollars. Several months income for an average family and not a priority purchase. For context, in 1954 the average house sold for $10,500 and a good car was $1,700.

According to the most recent data, 96.5% of households in the USA have a television.
(many have cut the cable and only consume on-demand streamed content, but they do so on a TV)
The numbers are similar for all developed countries; high nineties percentage ownership.
This is unsurprising given that TV has had a few 70 years to gain adoption.

Internet?

As of 2019, 89.4% of people in the USA have access to "broadband" internet. In 1995 when IDC first started collecting data on internet usage, 0.4% of the world population had access. In 24 years (less than a generation) proliferation of worldwide access has reached 58.4 % as of June 2019.

World Regions Population( 2019 Est.) Population% of World Internet Users30 June 2019 PenetrationRate (% Pop.) Growth2000-2019 InternetWorld %
Africa 1,320,038,716 17.1 % 521,614,944 39.5 % 11,454 % 11.6 %
Asia 4,241,972,790 55.0 % 2,275,469,859 53.6 % 1,891 % 50.5 %
Europe 829,173,007 10.7 % 727,559,682 87.7 % 592 % 16.1 %
Latin America / Caribbean 658,345,826 8.5 % 453,702,292 68.9 % 2,411 % 10.1 %
Middle East 258,356,867 3.3 % 175,502,589 67.9 % 5,243 % 3.9 %
North America 366,496,802 4.7 % 327,568,628 89.4 % 203 % 7.3 %
Oceania / Australia 41,839,201 0.5 % 28,636,278 68.4 % 276 % 0.6 %
WORLD TOTAL 7,716,223,209 100.0 % 4,510,054,272 58.4 % 1,149 % 100.0 %

See: https://www.internetworldstats.com/stats.htm

Smartphones?

81% of people in the US own and use a smartphone. The iPhone was released 12 years ago.
in 12 years we went from single digit percentage ownership to mass adoption.

This is only a proxy ...

The measures for consumer technology proliferation are only a relative proxy measure.
When it comes to the adoption of AI by companies, the "prize" (cost reduction / efficiency gains) is far more obvious to a business than to an individual. If the AI will save the trucking company $50k per truck per year and reduce accidents, that number can be fed into a spreadsheet and the investment can be analysed very fast. There will be a natural technology adoption curve with early adopters leading the way.

image

How Long Will It Take?

How long it will take for AI to proliferate and automate traditionally human jobs?
There is no consensus in the industry for how long it will take because the people with the most advanced AI are not advertising their progress. But Google, who are using AI for both their "assistant" and in many of their products including GMail response suggestion.

image

"Automation is going to further concentrate wealth in the hands of the few who own the robots" ~ Jane Kim (proposed a "Robot Tax" in SF)

Full Self Driving

The benchmark for AI/automation is "full self driving".

Elon Musk and his team at Tesla are predicting Model Y will be Level 5 Autonomous ("feature complete") by the end of 2020. This is just over a year from now.
image
https://youtu.be/knaskUXb12A

Even if this turns out to be another "optimistic" promise and it takes them a couple of years more, that's still only 3 years from now. Meanwhile Tesla, Waymo, Cruise, Uber (and many other smaller startups you haven't heard of) are racing towards level 5 autonomy.
Traditional auto-makers such as BMW, Volvo, VW, Ford and Toyota are all investing heavily in autonomous vehicle R&D to avoid being left behind, but only Tesla and Waymo have a fleet of vehicles on the public roads today.

Level 5 Autonomy will be achieved in the next 5 years without any doubt.
After that it's a regulatory/political issue; the politicians that decide to block the innovation by outlawing it from public roads may temporarily "protect" their constituents, but ultimately the technology will prevail because self-driving cars will save lives.
The countries/politicians that decide to embrace autonomous vehicles will reap the benefits of road safety and status for innovation and will shame the rest of the world into compliance.

The small but mighty nation of Singapore is leading the way in terms of adoption of fully autonomous public transport. They have trains and small vehicles in operation already:

image
https://www.smartnation.sg/what-is-smart-nation/initiatives/Transport/autonomous-vehicles

Several other countries are actively debating laws that are likely to pass including The Netherlands and Norway.
https://www.siliconrepublic.com/machines/countries-most-prepared-autonomous-vehicles-kpmg

Will your country allow autonomous vehicles in the next 10 years? Probably.
Which politician wants to be labeled a dinosaur by the voters?

What Can/Should We Do Next?

image

"The impact that technology has on society is not predetermined. It's a function of how we choose to respond to technological change." ~ [Michael Osborne] Professor of Machine Learning, Oxford University

Option 1: Burry our Heads in the Sand (Entertainment)

Most people are sleep-walking into a world where the relevance of humans is not guaranteed.
Play Fortnite (video/phone games), watch Netflix (TV) and/or scroll through Instagram and pretend that everything is going to stay the same forever. Then act surprised when the world is unrecognisable in 10 years time.

Option 2: Proactively Learn Everything about AI for Free Online

Learning about AI is available to anyone who has access to the internet.
Both MIT and Stanford have lectures and notes available on YouTube
and there are a plethora of tutorials anyone can follow.

As always, the choice is yours.

Further Reading

Relevant Videos (Aimed at Non-Technical Audience)

image
https://youtu.be/IHc5Zt7qT6o

Notes

1Time travelling economic migrants "people from the future" (derogatorily referred to as "Goobacks") because they go back in time to present day in order to do jobs paying pennies which, through the power of compound interest over hundreds of years, translate large sums for their families in the future.

Note: South Park is not for children.
https://southpark.cc.com/full-episodes/s08e06-goobacks
https://southpark.cc.com/full-episodes/s21e01-white-people-renovating-houses
https://southpark.cc.com/full-episodes/s22e09-unfulfilled

@nelsonic nelsonic added discuss Share your constructive thoughts on how to make progress with this issue in-progress An issue or pull request that is being worked on by the assigned person starter A beginner-friendly issue that is a good starting point for a new person labels Aug 30, 2019
@iteles
Copy link
Collaborator

iteles commented Aug 31, 2019

This is grim, but the logic is sound. The biggest questions here are around timelines and whether there will be enough non-AI jobs/roles left to support the burgeoning world population.

@nelsonic
Copy link
Owner Author

AI Weapons: https://youtu.be/GFD_Cgr2zho 😕

@nelsonic
Copy link
Owner Author

AlphaGo - The Movie | Full Documentary: https://youtu.be/WXuK6gekU1Y

@nelsonic
Copy link
Owner Author

https://youtu.be/K-dXlGIhKbw the info apocalypse 😕

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

nelsonic commented Aug 9, 2022

https://deephaven.io/blog/2022/08/08/AI-generated-blog-thumbnails/
via: https://news.ycombinator.com/item?id=32390526

"While the role of the artist isn’t going away soon, the role of stock image sites might disappear."

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

AI for the Next Era: https://youtu.be/WHoWGNQRXb0
image

@nelsonic
Copy link
Owner Author

DeepMind Flamingo is rapidly approaching AGI: https://youtu.be/zOU6usZRJvA

image

What a time to be alive! 🤖

@nelsonic
Copy link
Owner Author

Dall-E isn't that intelligent ... 🙄
"write EVILCORP in facebook font": https://labs.openai.com/e/6hnBpE7ST2JhpoRDQJdAhzlu
image

Perhaps I need to get better with my instructions/requests ... 🤷‍♂️

But a quick google reveals: https://www.google.com/search?q=facebook+font

image

i.e. Facebook has successfully boosted the SEO of some random page with a bizzare font in their AI Propaganda War.

Let's try again ...

image

Klavika

image

Clearly not going to replace a graphic designers in a hurry.

Good dicsussion on AI Art: https://news.ycombinator.com/item?id=33998112

@nelsonic
Copy link
Owner Author

NYT: ChatGPT is Code Red’ for Google’s Search Business
https://archive.ph/IxyHr
Via: https://news.ycombinator.com/item?id=34086462

@nelsonic
Copy link
Owner Author

Let's build GPT: from scratch, in code
https://youtu.be/kCc8FmEb1nY

@nelsonic
Copy link
Owner Author

ChatGPT plugins: https://openai.com/blog/chatgpt-plugins
How OpenAI will reach out into the private datasets ... 💭

via: https://news.ycombinator.com/item?id=35277677

@nelsonic
Copy link
Owner Author

image

image

image

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

"Pause Giant AI Experiments": https://youtu.be/8OpW5qboDDs

@nelsonic
Copy link
Owner Author

Ilya Sutskever (OpenAI Chief Scientist) - Building AGI, Alignment, Spies, Microsoft, & Enlightenment: https://youtu.be/Yf1o0TQzry8

@nelsonic
Copy link
Owner Author

nelsonic commented Apr 6, 2023

Sam Altman's Plans, Jobs & the Falling Cost of Intelligence: https://youtu.be/f3o1MW2G5Rs

@nelsonic
Copy link
Owner Author

Mid journey is incredible!! https://youtu.be/twKgWGmsBLY

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

The A.I. Dilemma: https://youtu.be/xoVJKj8lcNQ

@nelsonic
Copy link
Owner Author

The HUGE Problem with ChatGPT: https://youtu.be/l7tWoPk25yU
image

Clickbait-y title 🙄 but a good high-level overview with reference to AlphaGo 💭

@nelsonic
Copy link
Owner Author

Eric Schmidt on the Consequences of an A.I. Revolution: https://youtu.be/Sg3EchbCcA0

@nelsonic
Copy link
Owner Author

AI Riots vs "Moore's Law for Everything" by Sam Altman: https://youtu.be/JEFEJsTxPCc

@nelsonic
Copy link
Owner Author

nelsonic commented Mar 4, 2024

Racing towards the precipice … https://youtu.be/YZjmZFDx-pA

@nelsonic
Copy link
Owner Author

nelsonic commented Mar 4, 2024

@nelsonic
Copy link
Owner Author

nelsonic commented Mar 4, 2024

This is the tip of the iceberg: https://www.bbc.com/news/world-us-canada-68440150
image

This will be automated and Facebook will help Trump win with hyper-targeted advertising again. 🤦‍♂️

@nelsonic
Copy link
Owner Author

nelsonic commented Mar 8, 2024

The Dark Forrest: https://youtu.be/JrcbH0ge2WE

@nelsonic
Copy link
Owner Author

nelsonic commented Mar 9, 2024

Hard Takeoff Inevitable? https://youtu.be/71bSV-iHGLE

@nelsonic
Copy link
Owner Author

Dave Shapiro analysing Sam's motivations: https://youtu.be/hdBqgz_D2UI?t=900
image

the worst-case scenario for artificial intelligence is

"lights out for all of us".

The exact clip is: https://youtu.be/dXhoTrU1Kkw the "accidental misuse case" ...

@nelsonic nelsonic pinned this issue Mar 14, 2024
@nelsonic
Copy link
Owner Author

AI Agents Take the Wheel: Devin, SIMA, Figure 01 and The Future of Jobs: https://youtu.be/Dbog8Yw3kEM 💭

@nelsonic
Copy link
Owner Author

AI supremacy: The artificial intelligence battle between China, USA and Europe | DW Documentary
11K views · 3 hours ago

https://youtu.be/Le122vas9aM

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

nelsonic commented May 3, 2024

The Possibilities of AI [Entire Talk] - Sam Altman
https://youtu.be/GLKoDkbS1Cg

@nelsonic
Copy link
Owner Author

@nelsonic
Copy link
Owner Author

nelsonic commented Jun 5, 2024

Demis Hassabis shares what jobs he thinks are AI-proof: https://youtu.be/nwUARJeeplA scrub to 50:25
Human caring jobs basically.
We aren’t going to want robots/screens looking after our children … 🙃

@nelsonic
Copy link
Owner Author

How Ai Is About To Transform The World’s Economy: https://youtu.be/Aj9irFyFOxY

@nelsonic
Copy link
Owner Author

Preventing an AI-related catastrophe: https://80000hours.org/problem-profiles/artificial-intelligence/

@nelsonic
Copy link
Owner Author

“Ok Doomer”: https://youtu.be/ZXA2dmFxXmg
Clearly this becoming an “us vs. them” culture war fuelled by the interests that have the most to gain - short term - from “AI”. Meta …

@nelsonic
Copy link
Owner Author

nelsonic commented Sep 1, 2024

AI is here. What now? https://youtu.be/IZ4HOCld5nY

@nelsonic
Copy link
Owner Author

nelsonic commented Sep 1, 2024

The Potential for AI in Science and Mathematics - Terence Tao: https://youtu.be/_sTDSO74D8Q

@nelsonic

This comment was marked as outdated.

@nelsonic
Copy link
Owner Author

https://news.ycombinator.com/item?id=41638987

image

10 years later and image recognition is all open source and readily available: https://github.com/dwyl/image-classifier

@nelsonic
Copy link
Owner Author

nelsonic commented Oct 3, 2024

@nelsonic
Copy link
Owner Author

nelsonic commented Oct 3, 2024

@nelsonic
Copy link
Owner Author

nelsonic commented Oct 4, 2024

Career Advice For A World After AI: https://youtu.be/J4Hd5wudIrk

@nelsonic
Copy link
Owner Author

nelsonic commented Nov 9, 2024

Transformer: A Novel Neural Network Architecture for Language Understanding
https://research.google/blog/transformer-a-novel-neural-network-architecture-for-language-understanding/

Attention Is All You Need: https://arxiv.org/abs/1706.03762

@nelsonic
Copy link
Owner Author

AlphaFold3 is now open source: https://www.nature.com/articles/d41586-024-03708-4 💭

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
discuss Share your constructive thoughts on how to make progress with this issue in-progress An issue or pull request that is being worked on by the assigned person starter A beginner-friendly issue that is a good starting point for a new person
Projects
None yet
Development

No branches or pull requests

3 participants
@nelsonic @iteles and others