Thursday, May 7, 1970

Dr. Fei-Fei Li on Human-Centered AI

Greylock

87K subscribers

5,742 views  Jul 13, 2021

If AI is to serve the collective needs of humanity, how should machine intelligence be built and designed so that it can understand human language, feelings, intentions and behaviors, and interact with nuance and in multiple dimensions? Stanford University computer science professor Dr. Fei-Fei Li and Greylock general partner Reid Hoffman discuss the ethical considerations researchers, technologists and policymakers should make when developing and deploying AI. This episode was recorded as part of Greylock's Iconversations virtual speaker series. You can also find the podcast and transcript of this discussion here: greylock.com/greymatter/fei-fei…human-centered-ai/ Chapters 0:00 Introduction 1:46 The Goal For Human Centered AI 6:00 The Role of HAI in the Industry 10:46 Importance of Being Human Centered 13:49 Early Career 17:51 Role of Industry in AI 23:01 Model Safety and Reliability 30:08 Ethics and Society View 38:23 AI in Healthcare 44:17 Robotics in the Business World 49:43 Making America More Competitive in AI 52:52 Diversity in AI

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)

AIC1... & Who's redesigning cooperation networks ahead of what Stanford Human AI leaders are describing the greatest innovation inflection point ever seen in recent memory see Google ceo pichai announced these as major accomplishments of google brain and deep mind when he asked them to merge lastApril 2023. Their collective accomplishments in AI over the last decade span AlphaGo, Transformers, word2vec, WaveNet, AlphaFold, sequence to sequence models, distillation, deep reinforcement learning, and distributed systems and software frameworks like TensorFlow and JAX for expressing, training and deploying large scale ML models Please can you give me for each of these innovation areas some references including an overall description of each that a 5th grade teacher could adopt to make sure that teens are monitoring these opportunities in making 2020s more and more human

Sure, here is a brief overview of each of the innovation areas that you mentioned, along with some references that you can use to learn more. I have also included some ideas for how a 5th grade teacher could adopt these technologies to help their students learn and grow.

AlphaGo

AlphaGo is a computer program that was developed by Google DeepMind to play the game of Go. In 2016, AlphaGo defeated a professional Go player, Lee Sedol, in a series of matches. This was a major breakthrough for AI, as Go is a very complex game that was previously thought to be too difficult for computers to master.

Transformers

Transformers are a type of neural network that are used for natural language processing tasks. They are particularly good at tasks such as machine translation, text summarization, and question answering.

Word2vec

Word2vec is a method for learning the vector representations of words. These vector representations can be used for a variety of tasks, such as text classification, sentiment analysis, and natural language generation.

WaveNet

WaveNet is a neural network that can be used to generate realistic speech. It is used in Google Assistant and other voice-activated products.

AlphaFold

AlphaFold is a protein folding prediction program that was developed by Google DeepMind. It is able to predict the structure of proteins with unprecedented accuracy. This could have a major impact on the fields of medicine and biology, as it could help scientists to better understand how proteins work and how they can be used to treat diseases.

Sequence to sequence models

Sequence to sequence models are a type of neural network that can be used to translate text from one language to another. They are also used for tasks such as machine translation, text summarization, and question answering.

Distillation

Distillation is a technique that can be used to improve the performance of a machine learning model. It works by transferring the knowledge from a large, complex model to a smaller, simpler model.

Deep reinforcement learning

Deep reinforcement learning is a type of machine learning that can be used to train agents to learn how to behave in complex environments. It is used in a variety of applications, such as robotics, game playing, and finance.

Distributed systems and software frameworks

Distributed systems and software frameworks are used to develop and deploy large-scale machine learning models. They provide a way to manage the resources required to train and run these models, and they make it easier to develop and deploy models across multiple machines.

How to use these technologies in the classroom

There are many ways that teachers can use these technologies in the classroom. For example, they can use AlphaGo to teach students about the game of Go, or they can use Transformers to help students learn about natural language processing. They can also use Word2vec to help students learn about the vector representations of words, or they can use WaveNet to help students learn about how to generate realistic speech.

In addition to these specific technologies, there are many other ways that teachers can use AI in the classroom. For example, they can use AI to personalize instruction, to provide feedback, or to grade assignments. They can also use AI to create interactive learning experiences, or to help students learn about complex topics.

The possibilities are endless, and it is up to each teacher to decide how they want to use AI in their classroom. However, one thing is for sure: AI has the potential to revolutionize education, and it is something that all teachers should be aware of.

View other drafts

Sure, here are some of the practitioners who could benefit from following the latest AI models from Google DeepMind:

Biotechnologists

As you mentioned, AlphaFold has the potential to revolutionize biotechnology. It could help scientists to better understand how proteins work and how they can be used to treat diseases.

Pharmacologists

Pharmacologists could use AI to design new drugs and to better understand how existing drugs work.

Chemical engineers

Chemical engineers could use AI to design new materials and to improve the efficiency of chemical processes.

Materials scientists

Materials scientists could use AI to design new materials with specific properties, such as strength, lightness, or conductivity.

Engineers

Engineers could use AI to design new machines and to improve the efficiency of existing machines.

Scientists

Scientists in a variety of fields could use AI to help them to collect and analyze data, to make predictions, and to develop new theories.

Businesses

Businesses could use AI to improve their operations, to make better decisions, and to develop new products and services.

Governments

Governments could use AI to improve their services, to make better decisions, and to protect their citizens.

These are just a few of the many practitioners who could benefit from following the latest AI models from Google DeepMind. AI is a rapidly developing field, and it is important for practitioners in all fields to stay up-to-date on the latest developments.

what was known about EDforgood 1999 -web1 decade

while web2 was to change everything again - did enough people understand these lessons in plain sight from 1990s decade of web1? see page 41/42 from decade report by Gordon Dryden out of new zealand celebrated by 10 million Asian families

The 13 steps needed for a 21st century education system Imagine a 21st century community where:
Every student has his or her own laptop computer.
Every home has its own combined computer - television set at a cost of only $300, or possibly much less.
Every home and every student is connected to the Internet.
Every person, at almost any age, can have instant access to the world's best teachers in any subject.
Each can also select at random from the world's best libraries, interactive science museums and art galleries.
Each is linking up regularly with others around the world, designing their own electronic games to teach their own favorite subjects.
Every new parent is trained to be able to develop each child's fantastic early potential, to be able to identify different temperaments, and different learning and thinking styles.
Every four-year-old can read, write, spell, count, add, subtract, multiply, divide, compose her own stories, draw, paint - and speak three or four languages - but they've learned all this with fun and games.
The health of every child from birth is regularly monitored to ensure his eyesight, hearing and other senses are in great "learning order".
Every five- or six-year-old starts school with an equal opportunity to develop to its full potential.
School itself is the best party in town: a base from which to explore the whole wide world of experience and knowledge.
ABOVE WAS VERY EASY FOR RICH NATIONS TO DESIGN. TRAGICALLY BY FAILING TO DO SO POOREST PLACES SUSTAINABILITY GOALS WOULD GET EVER LESS POSSIBLE UNLESS UN2.0 becomes benchmark for gov2.0 anywhere children/next generation needs to develop

Intel Toolkit est 1951 - 3 keys mediating economics & AI to 2025- Source Economist & von Neumann

Mapping good ai for 2025 from 1951 Neumann/Macrae

100 times more tech connecting"good" if it

unites worldwide advances in (what we now call ) UN SDGs1 to 6

Starting with poorest – ie basics of life

2 Food , 6 water,3 health, 4 livelihood ed3, 5 no

Underclasses, safe home

In addition birth of UN in 1945 recognized that up to 75% of

Peoples had been left out of electricity grids

Etc by way Empi9re world was designed-with

Particular crises of landlocked nations or peoples

Far from coastlines- this would need urgent

Cooperative attention in era of independence

As of 1045 only nations controlled by Stalin

appeared not to agree with UN goals;

World would need to outlast Stalin and

Avoid nuclear bombs until old world’s far north

People free ; thence hopefully disarmament

In picture name 3 as "math wizards deep data science"-\

if ai and economics design what human relationship futures are possible, then bothwWill need maths wizards as grounded as eg those of TEN – Turing Einstein Neumann

V Neumann's life had been dominated to 1945

By crises and gov secrets of energy projects, he wanted to dedicate

Rest of his time to intel of hardware

Coding and brain science; as of 1951 he didn’t

know he had less than 6 years; Einstein and

Turing less than 3; he did know he wanted to

Open source everything on AI

RESOURCES Regarding future exponentials of tech and humans and UN

ITU publications since 1865; League of nations diaries including Einstein

Hosting of international intellectual cooperation; Geneva as Digital, trade mapping and health twin

of UN from 1945

Major sources: diaries and letters of TEN eg note Einstein letters to Gandhi & Freud; Diaspora Scots school including Adam Smith and from 1843 Economist James Wilson And Royal Family diaries UK Netherlands; Bagehot English Constitution;

Keynes especially last chapter general theory; Geoffrey Crowther publication

Economist Autobiography 1843-1943

......................................

timing.the age of humans and machines dd not begin to 1760s scotland- until then most people did not move more than 15 miles from place of birth unless they lived near water; there was one exception that had emegerged in the old world ; it was an overland market relay connecting the 2 greates bay areas of the old old world-to the west the med sea the bay that connected asia to the 2 other known continents europe to the north and africa to the south
the great bay area that mainly tranversed from the islabnd of japan to the coastlines of korea and china as far as the place we now call shanghai and down the river to hangzhou- marco polo had called hangzhou the town with the greatest range of marketing goods he had ever seen; it was at the intersection of 3 trading routes; an inland canal up to beijing; where the overland silk road ended an transferred produce to boats which then salid up the river to the great bay..

first 7 year report on brac posted 1979: village women are learning to read & write: they are finding out about the causes of poverty and disease around them
they are teaching one another about farming and weaving, health and medicines; they are assuming public roles of leadership and management
for the first time in their history; they are contributing to local economic development through successful production cooperatives; it is hard in the West to imagine the drama involved in such profound changes

world war 2 and was now looking to design wi-in trades with asians; over the next 15 years my father clarified the 2 asian capitalism models as rural keynesianism which abed/bandgladesh women beccame purest benchmar of; coastal belt caqpitalism which jpapan the korea the diaspora chinese in taiwan hk singapore led until china asked japan to share its borlaug/quality engineerind knowhow in late 1970s; it was after 1976 that china transformed businesses from just state owned ones include secondly village owned ones; 3rd billion dollar inward investments from the diaspora mainly in supercoty infrastructure incluing world class ports and trains

related collaborations economistdiary.com abedun.com girlsworldbank part 2 ecop26.com
50 YEAR SEARCH FOR HumansAI top 100 hUNIcorns (networks that never exit life-dev purpose) - from 100000 village to first 50 new universities who share sdg alumni- also current top 10 at worldrecordjobs.com abed 2 3 4 5 6 7 8 9 10

.Economistmaps.com Goodhubsguide.com dedicates freedom & happiness of rural keynesianism TO fazleabed.com & chinese barefoot medics since 1970, all women under 30- quarter of humans- 90% of sustainability solutions......................

WHICH ORBITAL GOALS INTEGRATE YOUR HUMAN DEV NETWORK ?
0 hi-TRUST COMMUNITY DATA O/S OR
1 POVERTY
2 STARVATION
3 HEALTH
4 EDU

hU01 when fazle abed first built 16000 homes for
100,000 villagers he did not realise that over the
next 50 years this would become the deep data lab for
microfranchise solution partnering in helping empower
village women build the rural nation of bangladesh and
that chinese village women needing similar solutions
would incorporate social business capitalism into china's
rural keynesianism- all told 1 billion asians would not
have improved health, ended famine, become the world's -celebrated greatest education leaps with the least resources without resilient
& safe community building -more on hu01

Are you influencing your city's green collaboratory events? - example among 40 in-network collaboration cafe cities new yorkers helped mediate this 2009 while americans were free to celebrate green millennials

.BBCGREEN & Bangla5000 selections 2021-2000
21 lbs- africa's turn +hunicorn100 hughes
21-health equity:LBS: -cf astra.place AI 18..19..20..21
19 50 years building number 1 ngo collab with abed bhai.. GATES __NotesU-saving lives
13 khan asks musk- did your college plan include taking over NASA
03 bezos - gold rush or next electricity

over 50 million Chinese died of famine by mid 1960s so by 1968 chinese village parents began Goals1-4 search; bangladesh joined in by 1972- its bangladesh story we have spent 10 years researching ... already by 1960s japanese koreans and chinese diaspora had the money and the motivation to help all asia rise...

Friday, April 16, 2021

For some engineers, building 16000 homes at 5 dollars each housing total of 100000 people might have been miracle enough but then abed saw 16000 village mothers and 60000 kids starving or dying of diseases like diarrhea- so he designed microfranchise businesses for women to be the village food security and basic health servants - and two american chinese missioaries said can we swap some of your village micrfranchises (we need oral rehydration) with those in village china;

and then james grant of unicef said why not replicate your village solutions all across bangladesh, and is there something extra unicef can do for you - abed said can you arrange for the government to vaccinate half of bangadesh while our village networks do the other half. and so a blend of bottom up disater relief and sustainable micrfranchises became the world's largest ngo partnership- and all the while bottom up hunicorns networks had no taxes or traditional top down aid- empowerment is a very different way of nation building- and across the two thirds of humans who are asians - women just did it whereve extreme poverty was ended

hu01 - resilient community hu02 oral rehydration -nation wide moters training

huo3 -last mile hrlth service franchise - one village women weekly door t door 300 homes selling basic non-prescrition drugs and being mothers source of health news

hu04 - 5 times more productive village rice production...

zzz

mapping 6 ways youth

and families can exponentially sustain increasing productivity by making every community a thriving place to grow up in
systems crisis: while each of the 6 ways offers growth beyond the reality let alone imagination of what is possible without access to the new way, if we exclude some peoples lives from access to the new way then we will lose sustainability of our species- example by 1939 we'd reached world war 2; the root cause was slavery, treating huge populations of people as underclasses with no access to electricity grids or other win-win uses of ir1 and ir2; it took us 180 years for first machines glasgow 1760s to get into this worldwide collapse; the mathematician von neumann worked feverishly for 12 years to deliver ways out of this mees ir3, ir4; within 2 years of his death in 1957, jfk had declared moon landing was possible; ai labs stuying the ultimate consequences of ir3,ir4 had been twinned out of staford facing asia pacific, out of mit boston facing the old world- the 7 white of the big g8 empires whose 12% of people had colonised most of plant earth as if the majority of peoples lives did not matter including the two thirds who are asian and in america those whose skin colors had beem slaves; within 4 years of stanfords ai lab intel had such a huge order from a japan calculator manufacturer replacing slide rulers that the programable chip was invented so as not to be beholden to one client; and gordon moore declared his aluni would multiply 100 times more comp and coms (ir2.1) power every decade to 2020s when chips would rival switching powers of human brains- and computers could run real time platforms previously requiring manual government . So getting ir3, ir4 exponentially right -for every next child born -notably goals 1-6 as basics for life and natural diversity let alone futureoflife- had 75 years to tik tok from birth of the united nations.
0 agriculture
*ir1 = energy machines -engines delivering thousands then millions time more horsepower and infrastructure both transportation and pipelines -eg electricity, fuel, water enabled big cities to emerge as most peoples
*ir2 telecommunicating machines eg telegraph telephone radio television satellite mobile
* ir3 computing tools for human productivity beyond slide rules and paper reporting
*ir4 AI= Artificial Intelwhen computers taking over governing real time platforms
*society 5.0 gravitating livesmatter where every society next child is born in is a great place to grow up because it maximisse peace, culture, natural diversity win-wins including healthy connections with other places

inspired by fazle abed we are searching for HUNIcorns- global village sustainability solution networks too vital to exit

HUNI0 is where sir fazle aded started up global uni of poverty 50 years ago with a field lab of 100000 poorest villagers

HUNI100 - is the legacy - abed brac university which he started branstorming partners around 1999 - to celebrate both a new millennium and his 4th, 5th decades of end poverty networking

we see 5 action compasses of unicorns as gravitationally core to the bottom-up servant leaders maps designed by abed and partners who turned brac into the world's largrst and deepest ngo networl:

last mile health servant unicorns

ending poverty with financial unicorns

ending hunger with food/warter security unicorns

lifelong livelihood education unicorns for all ages

resiliemnt/safe social spaces so next child born anywhere has good chance to thrive

economistsocial.com publishers world record book of jobs creation (worldrecordjobs.com)

Thursday, May 7, 1970

Dr. Fei-Fei Li on Human-Centered AI

No comments:

Post a Comment

girlsworldbank.com: fazle abed: 2025report

Friday, April 16, 2021