AI NEWS OpenAI vs Helen Toner. Is 'AI safety' becoming an EA cult?

Summary

The video discusses recent controversies surrounding OpenAI, focusing on the dismissal of Sam Altman and the subsequent fallout. It examines claims made by former board member Helen Toner, who alleges being kept in the dark about AI developments and accuses Altman of a history of deceit. The video also critiques the effective altruist movement's influence on AI safety, highlighting their extreme views on halting AI progress and the potential for global surveillance. The narrative questions the motives behind these actions and urges viewers to consider the broader implications of letting a vocal minority dictate AI regulation.

Takeaways

• 🗣️ A former OpenAI board member, Helen Toner, has spoken out about the circumstances surrounding Sam Altman's firing, sparking controversy and debate within the AI community.
• 🔍 Helen Toner claimed that she and others were kept in the dark about significant developments at OpenAI, such as the launch of Chat GPT, which they only learned about through Twitter.
• 🚫 OpenAI's current board has refuted Helen Toner's claims, stating that they commissioned an external review which found no evidence of safety concerns leading to Sam Altman's departure.
• 👥 The debate has become somewhat tribal, with people taking sides and supporting the narratives that align with their pre-existing views rather than objectively assessing the situation.
• 💡 There are concerns that the conversation around AI safety is being dominated by a minority with extreme views, potentially skewing the direction of AI regulation and research.
• 🌐 Some individuals within the effective altruist movement are pushing for stringent global regulations on AI development, including bans on certain technologies and surveillance measures.
• 🕊️ The term 'AI safety' has been co-opted by groups with apocalyptic views on AI, leading to confusion and a tarnishing of the term for those working on legitimate safety concerns.
• 💥 There is a risk that the focus on existential risks from AI could overshadow more immediate and tangible concerns about AI's impact on society and the need for practical safety measures.
• 📉 The influence of certain organizations and individuals with extreme views could have negative repercussions on the AI industry, potentially stifling innovation and progress.
• 🌟 The video script emphasizes the importance of balanced and evidence-based discussions around AI development and safety, rather than succumbing to fear-mongering or cult-like ideologies.

Q & A

What is the main controversy discussed in the video script?

The main controversy discussed is the dismissal of Sam Altman from OpenAI and the subsequent claims and counterclaims made by various parties, including Helen Toner, an ex-board member, and the current OpenAI board.

What was Helen Toner's claim about the Chad GPT revelation?

Helen Toner claimed that she and the board learned about Chad GPT on Twitter, suggesting they were kept in the dark about this significant AI breakthrough.

How did OpenAI respond to Helen Toner's claims?

OpenAI responded by stating they do not accept the claims made by Helen Toner and another board member. They commissioned an external review by a prestigious law firm, which found that the prior board's decision did not arise from product safety or security concerns.

What is the significance of GPT 3.5 in the context of the video?

GPT 3.5 is an existing AI model that was available for more than 8 months before the release of Chat GPT. It signifies that the technology behind Chat GPT was not new, but its user interface and format as a chat application became popular.

What was the claim made by Helen Toner about Sam Altman's past?

Helen Toner claimed that Sam Altman had a history of being fired for deceitful and chaotic behavior, including from Y Combinator and his original startup, Loopt.

How did Paul Graham, the founder of Y Combinator, respond to the claim about Sam Altman's dismissal from Y Combinator?

Paul Graham clarified that Sam Altman was not fired but rather agreed to step down from Y Combinator to focus on OpenAI when it announced its for-profit subsidiary, which Sam was going to lead.

What is the concern regarding the influence of the Effective Altruism (EA) movement on AI policy?

The concern is that the EA movement, with its belief in the imminent risk of AI superintelligence and potential existential threats, may be pushing for extreme regulatory measures that could stifle innovation and progress in AI.

What is the view of some researchers and experts on the existential risk posed by AI?

Some researchers and experts believe that while existential risks could emerge, there is currently little evidence to suggest that future AIs will cause such destruction, and more pressing, real-world concerns about AI should be addressed.

What is the criticism of the EA movement's approach to AI safety?

The criticism is that the EA movement has hijacked the term 'AI safety' and focuses on extreme doomsday scenarios, which overshadows more practical and grounded concerns about AI's impact on society and the need for sensible regulations.

What is the argument made by the video script against the extreme regulatory measures proposed by some AI safety advocates?

The argument is that extreme measures, such as global bans on AI training runs or surveillance on GPUs, are not rational and could have disastrous consequences, such as nuclear conflict, which should not be the basis for governing and regulating AI development.

Outlines

00:00

🤖 AI Controversy and Board Member's Claims

The video delves into the controversy surrounding the dismissal of Sam Altman from OpenAI, with a focus on the claims made by Helen Toner, an ex-board member. It discusses the community's divided opinion and the 'bombshell' revelations from Toner's interview, such as the allegation that the board was kept in the dark about 'Chat GPT' until it was revealed on Twitter. The video also mentions the response from the current OpenAI board, which refutes Toner's claims and highlights an external review conducted by the law firm WilmerHale that found no evidence of AI safety concerns leading to Altman's departure.

05:00

🔍 Misrepresentations and the Reality of AI Developments

This paragraph addresses perceived misrepresentations made by Helen Toner regarding the OpenAI situation. It clarifies that the technology behind 'Chat GPT' was not a secret and had been available for months, suggesting that Toner's claim of being informed about it on Twitter might be an exaggeration. The paragraph also refutes the claim that Sam Altman was fired from Y Combinator for deceitful behavior, with Paul Graham, a mentor to Altman, clarifying that Altman's move was a mutual decision to focus on OpenAI rather than a dismissal.

10:01

🧐 The Influence of Effective Altruism on AI Policy

The video script discusses the influence of the Effective Altruism (EA) movement on AI policy, suggesting that some members have extreme views on AI safety, such as the belief in an imminent AI superintelligence that could lead to humanity's extinction. It raises concerns about the EA's approach to AI regulation, which includes ideas like banning certain hardware and enforcing global surveillance to prevent AI development. The paragraph also highlights the potential negative impact of these beliefs on the broader conversation around AI safety and policy.

15:02

💡 The Distortion of AI Safety Discourse

This section of the script criticizes the distortion of the AI safety discourse by certain groups with extreme views on AI, which it associates with the Effective Altruism movement. It argues that these groups are overshadowing more grounded and pressing concerns about AI's real-world applications and potential harms, such as the impact on marginalized communities. The video calls for a more balanced and evidence-based approach to AI safety, rather than one driven by fear of an AI apocalypse.

20:02

🛑 The Risks of Overzealous AI Regulation

The speaker expresses concern over the potential risks associated with overzealous AI regulation, particularly that advocated by certain groups within the Effective Altruism movement. The paragraph outlines extreme regulatory measures such as making hardware illegal or imposing pervasive surveillance on data centers. It emphasizes the need for a more nuanced and practical approach to AI regulation that doesn't stifle innovation and progress.

25:02

🌐 The Importance of Openness in AI Development

In this paragraph, the script highlights the benefits of open-source AI and the importance of sharing knowledge to improve security and prevent vulnerabilities, drawing a parallel with cybersecurity. It contrasts this with the views of a minority advocating for extreme measures such as global surveillance and potential military conflict to halt AI development. The speaker argues against letting such extreme perspectives govern and regulate AI, advocating for a balanced and rational approach to its development and safety.

🚀 Balancing Progress and Caution in AI Development

The final paragraph wraps up the discussion by emphasizing the need to balance progress and caution in AI development. It criticizes the extreme views that suggest halting all AI progress and instead calls for a reasoned, evidence-based approach to managing risks. The speaker encourages viewers to consider multiple perspectives on AI safety and to be wary of letting cultish ideologies dictate the future of AI regulation and development.

Keywords

💡Open AI

Open AI is a research laboratory that focuses on developing artificial intelligence (AI) technologies. In the context of the video, it's the organization where significant events have transpired, including the departure of Sam Altman and the subsequent controversy. The script discusses the internal conflicts and the board's response to the events that unfolded at Open AI.

💡Sam Altman

Sam Altman is a prominent figure in the tech industry and was a key player at Open AI. The video script revolves around his departure from Open AI and the ensuing debate over the circumstances of his firing. His character and actions are central to the narrative being discussed.

💡Helen Toner

Helen Toner is mentioned as an ex-board member of Open AI who has spoken out about the events surrounding Sam Altman's departure. Her perspective is presented as one of the key sources of information about the internal happenings at Open AI, and the script discusses her claims and their reception.

💡AI Regulation

AI Regulation is a central theme in the video, which discusses the need for oversight and control over AI development. The script references Eliza Yosi's views on regulating AI, suggesting an international arrangement to monitor and license AI training technologies, reflecting on the broader conversation about how to manage the progress and safety of AI.

💡AGI (Artificial General Intelligence)

AGI refers to a form of AI that possesses the ability to understand, learn, and apply knowledge across a broad range of tasks at a level equal to or beyond that of a human. In the script, the concern about AGI's potential risks and the need to manage its development are highlighted, particularly in relation to the discussion on AI safety and regulation.

💡Effective Altruism

Effective Altruism is a philosophical and social movement that uses evidence and reason to determine the most effective ways to benefit others. The script mentions this movement in relation to certain individuals and organizations that have a significant influence on AI policy and the narrative around AI safety, suggesting a potential bias towards apocalyptic AI scenarios.

💡AI Safety

AI Safety is a field of study focused on ensuring that AI development is conducted in a manner that avoids potential harm to humanity. The video discusses how some groups have co-opted the term to push an agenda that may not be representative of the broader AI community, leading to a distorted view of AI risks.

💡EA (Existential AI Risks)

EA, in the context of the video, refers to groups concerned with existential risks posed by AI, which are risks that could lead to the extinction or irreversible degradation of humanity. The script critiques this perspective as being overly alarmist and potentially detrimental to a balanced conversation on AI development and policy.

💡WilmerHale

WilmerHale is a prestigious law firm mentioned in the script as having been commissioned to conduct an external review of the events leading up to Sam Altman's resignation from Open AI. Their findings are significant as they contradict some of the claims made by Helen Toner and others.

💡Chad GPT

Chad GPT is a term used in the script to refer to a demo or iteration of AI technology that gained significant attention. The script discusses the controversy around its release and the claims that board members were kept in the dark about it, which is a point of contention in the narrative.

💡Y Combinator

Y Combinator is a well-known startup accelerator that has been influential in the tech industry. The script references Sam Altman's history with Y Combinator, specifically addressing claims about his departure and the actual circumstances that led to him focusing on Open AI instead.

Highlights

Ex-board member of OpenAI, Helen Toner, speaks out on Sam Altman's firing.

OpenAI's response to Helen Toner's claims, denying the allegations made by her.

External review by law firm WilmerHale found no AI safety concerns led to Altman's departure.

Chad GPT's release was not a secret, contrary to Toner's claims; it was based on existing GPT 3.5 technology.

Paul Graham clarifies Sam Altman was not fired from Y Combinator, contradicting media narratives.

Critique of Helen Toner's actions and EA (Effective Altruism) movement's influence on AI policy.

Concerns about EA's extreme views on AI regulation, including global bans and surveillance.

EA's shift from humanitarian goals to focusing on AI doomsday scenarios.

Debate on the validity of EA's claims about AI's potential to cause human extinction.

Impact of EA-funded research on the AI policy landscape in Washington.

Criticism of EA's approach to AI safety versus other researchers with different focuses.

Discussion on the need for balanced AI regulation that accounts for both risks and benefits.

The importance of not letting extreme views dominate AI development and policy.

EA's self-identification as part of AI safety and the confusion it causes in the field.

The potential consequences of EA's influence on global AI development and international relations.

The contrast between EA's apocalyptic views and more nuanced perspectives on AI's future.

The call for a rational and balanced approach to AI development and its associated risks.

Transcripts

big AI news today so an ex-board member

of open AI comes out talking about what

actually happened during Sam alman's

firing we covered that a few days ago

her name is Helen toner but today we get

the opening eye response as well as some

other people that weigh in on whether or

not some of the claims that she makes

are truthful or not so let's take a look

at that but before we do take a listen

to this 30 second clip of Eliza Yosi and

tell me do you agree with what he's

saying do you agree with how he thinks

we should regulate AI how you feel about

what you will hear today will greatly

depend on whether you agree with what he

is saying should this type of research

and development be made against the law

yeah basically I think that we should

track all the gpus have international

arrangements for all of the AI all the a

training Tech to end up in only

monitored supervised licensed data

centers and Allied countries and and

just like not permit training runs more

powerful than GPT 4 like that whole line

of reasoning is not that regulating the

stuff will protect you from a super

intelligence because it will not that's

more in the hopes that people change

their minds later maybe after some major

disaster that doesn't kill everyone you

don't like press the off switch to deal

with the super intelligence the super

intelligence does not let you know that

you need to press the off switch until

you are already dead AGI rolls around

only once subscribe so Helen toner comes

out spilling the beans on what happens

during that whole opening ey Fiasco

where Sam Alin got fired tons of people

had to be brought in to kind of manage

the situation and figure out who's going

to be running open AI moving forward one

thing that's obvious to me is that the

community the people following this you

I everybody else were very divided on

what actually happened who is at fault

who is to blame who's telling the truth

who's being honest so the kind of big

bombshell revelations in Helen toner's

interview were the following the biggest

one that I think a lot of people quoting

is the fact that they learned about Chad

GPT on Twitter that was a line that she

used to kind of point out that they were

being kept in the dark about chadt this

great new breakthrough in AI technology

also she mentioned that Sam had a

history of being fired for his deceitful

and chaotic Behavior as she puts it

maybe she was quoting somebody but she

was saying he was fired from why

combinator he was fired from looped his

original startup and also that Sam

didn't inform the board that he owned

the open AI startup fund now we've

covered that interview already but since

then there's been a lot more sort of

Revelations about what's actually been

happening some people had a chance to

kind of reply back to some of these

allegations so today let's take a look

at it now unfortunately I know that a

lot of people this seems to be becoming

kind of a tribal thing right so you have

red team blue team and you're just

rooting for the people that you want to

win I don't know if that's the right

approach here I think as you'll find out

there's a lot more gray area here than

at first meets die but let's take a look

so first of all all here is the opening

ey board the current one responds to

Helen toner about her claims they're

saying we do not accept the claims made

by Miss toner and Miss mcau regarding

events at open AI so those were the two

board members that were kicked off after

the firing of Sam Albin so they're

saying the first step we took was to

commission an external review of events

leading up to Mr Altman's forced

resignation and this is true they hired

a as they call a prestigious Law Firm

we've covered this when this happened

Wilmer hail they led the review they

conducted dozens of interview with

members of open ey's previous board

including Miss Stoner and Miss mcau

openi executive advisers to the previous

board and other pertinent Witnesses

reviewed more than 300,000 documents and

evaluate various corporate actions and

both mm mcau provided ample input to the

review there you find it rejected the

idea that any kind of AI safety concern

necessitated Mr alman's replacement and

that legal company found that the prior

board's decision did not arise out of

concerns regarding product safety or

security the pace of development opening

eyes finances or its statements to

investors customers or business partners

they say we regret that Miss toner

continues to revisit the issues that

were thoroughly examined by Wilmer hail

LED review rather than moving forward m

t has continued to make claims in the

Press although perhaps difficult to

remember now openi released chpt in

November 2022 as a research project to

learn more about how useful its models

are in conversational settings it was

built on GPT 3.5 an existing AA model

which has already been available for

more than 8 months at the time this is

an important thing to understand that

this technology was available everyone

knew about it there were companies like

for example Jarvis AI I think was called

and then I think they changed their name

to Jasper AI but it was basically for

writing SEO optimized articles they were

running on GPT 3.5 for a long long time

if you look at open ai's YouTube channel

this was 2 years ago so this was August

10th

2021 that seems like a lot more than two

years ago so this is the open AI codex

that ilas sover and Greg are presenting

here which already resembles a back and

forth kind of chat application but this

one is for coding this was codex and

they're inviting people to participate

to use the API to use it for their own

needs so there wasn't really anything

new that was released Chad GPT was a

little demo nobody thought it was going

to blow up the way it did they took an

already existing technology that

hopefully everyone was aware of if

you're in the board I hope you were

aware what the company was working on

and they just packaged it up in a chat

format and they put out there for

research purposes and it blew up again I

don't think anyone quite expected that

it would become the fastest growing

apple of all time do you think that

could have been possible to just predict

that would happen I remember Elon Musk

posting that it's scary good and I think

that's what got me to try it initially

so her saying that she was made aware of

Chad PT on Twitter

not exactly sure what that means either

she just wasn't paying attention to what

was happening in the company cuz again

all those pieces of Technology were

available they were available to the

public there was the playground there

was the API the only thing that changed

was the user interface and that user

interface just clicked with everybody it

opened up everyone's eyes to what was

possible so I think it's one of those

things where she's seeing the truth but

the reason people are reposting it

because it sounds so much scarier like

she had no idea this product was

released it was already released it was

just a different UI that was kind of put

in place I think that's a fair thing to

say because again the playground the web

page where you can mess around with it

that was already available API was

already available so that's Point number

one that she made that to me seems like

a dubious claim it's either made to

sound Sensational when it's it's really

not I'm not sure but it just something

about it feels a little bit fishy but

let's continue she also said that Sam

Alman was fired from a y combinator for

his deceptive and chaotic Behavior So

Yesterday tons of newspapers popped up

with their stories about how Paul Graham

fired Sam malman from y combinator so

Graham who was something of a mentor to

the young Tech Guru to Sam Alman flew

from the United Kingdom to San Francisco

to personally give his prote the boot to

fire him here's the Washington Post

albin's polarizing past hints at opening

eyes board's reason for firing him so it

says same thing that Graham flew from

United Kingdom to San Francisco to give

his prote the boot he fired Sam Alman

before we continue can we agree that

that's what these newspapers say right

this the impression that they give you

if you had to summarize it to somebody

would that be an accurate summarization

of what it says here I think so right

here's the problem with that Paul Graham

today commented on what actually

happened he said I get tired of hearing

that white commentator fired Sam so

here's what actually happened for

several years he was running both YC and

openi but when open announced that it

was going to have a for-profit

subsidiary and that Sam was going to be

the CEO we specifically Jessica told him

that if he was going to work full-time

on openi we should find someone else to

run white combinator and he agreed if he

said that he was going to find someone

else to the be the CEO open thei so that

he could focus 100% of white combinator

we would be just fine with that too we

didn't want him to leave just to choose

one or the other now some people in the

comments are trying to push this to say

that well that's what firing is right

that's the same thing and Paul responds

no we would have been happy if he stayed

and got someone else to run open AI

can't you read I mean if you are okay

with a person staying and running your

company if the issue is you just don't

want to have a split Focus or

potentially conflict of interest you're

saying hey either you're 100% here or

100% there the fact that you're okay

with them being 100% running your

company that's not the same as firing

that's not the same as you know as the

Washington Post puts it giving him the

boot how are you want to phrase that

that's not what happened here so again

that to me seems like another lie that

everybody bought from Helen toner and

again as we covered yesterday the issue

here is that some of the organizations

that she is somehow affiliated with do

specifically tell the members you know

here are our talking points here's what

we actually believe right on one hand

and on other hand here's our talking

points for the normies for the people

that might not agree with us here's a

post by Nathan Lans saying Paul Graham

says a story about Sam Alton being fired

from White commentator is not true I

think there are many cases like this

where people are assuming things bad

about Sam that probably aren't true I've

never met Sam but I've only ever heard

great things about him as a person most

say he's one of the most genuinely nice

and intelligent people they've ever met

which is again before a lot of the stuff

that was happening with open AI you

would only hear nice things about Sam

Alman by the way me personally I'm never

surprised that some of these high

charging people aren't quot unquote nice

I think deep down to go that hard after

some of these goals you kind of have to

be a bit of a killer look at Steve Jobs

right when his biography was released

and we started learning about how he was

and I mean there were rumors about

beforehand but like he wasn't a very

nice guy people didn't really love him

all the time in fact some of the people

that work closely with him afterwards

said that while they really didn't like

working with him because of how he was

later on they reflected saying that

during that time they were pushed harder

and they accomplished more like they

just extracted more from themselves they

the the output was much better much

higher level because of how he pushed

them same thing with Elon Musk right the

latest book that was released about him

talks about demon mode this rage that he

goes into to push certain projects

through to put pressure on people Bill

Gates was a nice guy we're always so

surprised when these high charging

people that achieve so much Against All

Odds aren't super duper nice ell in the

general

remember when it came out that maybe she

wasn't the greatest person to her staff

or whatever and people were shocked but

she dances so well she always does her a

little happy dance how could she be a

bad person so none of this is to say

that any of this is defending Sam Alman

necessarily not saying he's the saint

and anybody against them are the bad

guys and you're free to not like Sam

Alman and disagree with how he's running

things that's not the point of this

video to dissuade you from that I'm just

saying don't buy into everything that's

said especially by people that might

have very specific motives that they

want to push through that might be hard

to convince people to follow you if you

actually save what your motives are

right and we'll talk more about that in

a second here's Rick Burton so I'm not

familiar with this person so take this

with a grain of salt but he's somebody

that came out and spoke against Helen

toner saying I lived in a community with

Helen toner let me tell you what she is

Helen is the very worst that Academia

has to offer she thinks opinions matter

more than actions while she was writing

puff pieces about China Sam Alman was

working she lucked into the open AI

board and Stage a coupe Helen toner has

destroyed value she has created nothing

of value she's not open to open debate

and now she's using her dying voice to

hurt Sam Helen completely misunderstood

what a board does from day one it is

therefore quarterly oversight and acting

as a check on the CEO she never gave a

feedback to Sam she just tried to fire

him this is not what a competent board

does they work on the problems again as

she talks about in the interview they

assumed that Sam if he heard about the

firing if if he got any word about it he

would trying to counter acted somehow so

they purposely set up it in such a way

that he would not have any knowledge of

it they didn't warn him they didn't talk

to him they as far as I can tell never

tried to work the problem out also again

if a board there is for quarterly

oversight then yeah they're probably not

reporting to her every UI change or

every release of an app she must have

known about GPT 3.5 or gpt3 before it

because it was there 8 to 10 months

before the release of Chad GPT which

again was more of a UI change I know it

seems big to us now but the technology

was all there it just the rapper that

got released to the public was much more

popular than anyone could have expected

now one of the reasons that people are

worried about people from a background

in EA the effective altruist

organization if they're somehow linked

to it you know here for example a lot of

people at the nist the staffers would

revolt against expected appointment of

an EA AI researcher the reason being is

that a lot of the beliefs that these

organizations have are probably not the

beliefs that you and I share things they

believe are things like that we're only

months or years away from building an AI

superintelligence able to outsmart the

world's Collective efforts to control it

right so we're potentially months away

from having an AI super intelligence and

what does that mean well according to

Elijah yudkowsky if stopping malignant

AI requires war between nuclear armed

Nations that would be a price worth

paying do you agree with that mentality

like should we pause all AI indefinitely

stop any progress try to control all the

gpus so that nobody's able to research

or or do any work with AI and then spy

on other nations to make sure they're

not doing it and if they are then

nuclear war is an option to shut them

down to prevent them from working on AI

do you agree with that statement this is

political.com by the way that I'm

reading from it continues the profits of

the AI apocalypse are boosted by an

avalanche of tech dollars and also

billions in crypto funds is we'll see in

a second with much of it flowing through

open philanthropy a major funer of

effective altruist causes it's an epic

infiltration said one biocurity

researcher in Washington right and a lot

of these EA people the members of this

movement so they're usually white

typically male and often hail from

privileged backgrounds and like many of

her peers conell calls EA a cult right

Sam bman freed is part of that as some

would say cult right he's convicted for

stealing as much as 10 billion from his

customers they literally believe that

they're saving the world that's their

mission these effective altruists truly

believe what they're saying about AI

safety the idea that within a few months

or a few years it'll cause the

extinction of the human race unless we

stop all progress on it now and this is

the problem with all of that and this is

the thing that I hope more people

understand is that they refer to

themselves as being part of AI safety

they've kind of hijacked that term

instead of saying they're ai doomers ai

apocalypse you know Terminator 2 AI is

going to turn us all into paper clips

they say AI safety which is a problem

because as many longtime Ai and

biocurity researchers in Washington say

there's much more evidence backing up

their less than existential AI concerns

While most acknowledge the possibility

that existential risks could one day

emerge they say there's so far little

evidence is there any evidence but

they're saying there's little evidence

to suggest that future AIS will cause

such a destruction even when paired with

biotechnology the point here is that we

have much more pressing concerns than

Terminator robots marching down a street

yes we need to be cognizant of that yes

we need to make sure we're not ignoring

the risk of Rogue AI Etc but as AI is

coming into the various software various

businesses various government

organizations we have very pressing very

real concerns about how it's going to be

used and these doomsday scenarios are

corrupting that conversation they're

leading us away from actual things that

matter it's like imagine if you're

trying to regulate Cars and auto safety

things like seat belts making sure that

you have enough lights you know crumple

zones Etc like the safety features of

automobiles but there was this other

group that was slowly taking over

Washington that actually defined Auto

Safety car safety as no we have to get

rid of all cars forever because cars are

dangerous so while the normal reasonable

people are trying to make cars safer the

other people are saying we also want to

make cars safer but their goal is

actually just getting rid of all of it

because they have this belief that one

day cars will rise up and kill everybody

or whatever you probably don't want

those people making the regulations

here's an example where Deborah Raji a

Mozilla fellow in AI researcher at

Berkeley you know her research focused

on how AI can harm marginalized

communities but that was completely

overshadowed by open philanthropy funded

researchers that suggested that llms

like Chad GPT could supercharge the

development of bioweapons and so Raji

saying well if you just look online for

a second you can find all that stuff on

Google the fact that you can get the llm

to regurgitate that stuff if you try

hard enough there's nothing exceptional

about it but her research is left in the

dust because she doesn't have the funds

that these other researchers have that

are concerned about an AI doomsday

scenario as EAS bring their message to

virtually every corter of the nation's

capital experts are warning that the

tech funded flood is reshaping

Washington's policy landscape driving

researchers across many organizations to

focus on the existential risks posed by

new technologies often to the exclusion

of other issues with firmer empirical

grounding as another researcher puts it

I don't want to call myself AI safety

that word is Tainted now right they want

to call themselves something different

like system Safety Systems engineer

right because saying AI safety now might

link them to this cult-like organization

concerned about things that again don't

really have too much proof behind them

by the way the founders of anthropic

used to consider themselves part of EA

now it seems like maybe they're kind of

trying to distance themselves away from

that movement here's stepen Pinker

cognitive scientist at Harvard saying

it's upsetting to read how the shift in

EA from saving lives in Africa because

again that's where they started trying

to do good for the world for Humanity

reduce poverty reduce suffering Etc so

from saving lives in Africa to paying

brainiacs to fret about how AI will turn

us into paperclips may not have been

Mission Drift But bait and switch EA's

core ideas still sound and some some of

its Charities are praiseworthy I hope

the movement regains its bearings here's

yakun saying as I have pointed out

before Aeris is a kind of apocalyptic

cult why would its most vocal Advocates

come from Ultra religious families that

they broke away from because of Science

and the big concern with some of the

people that believe in this stuff if

they're allowed to regulate where we're

going with AI their goals aren't to

create Common Sense regulations like

would most of us are talking about

having some sort of visibility into

who's building what having some sort of

a reporting requirements some safety

limits some safety testing red teaming

efforts Etc right that's that they're

saying that they want Common Sense

regulations that's what they're telling

you but what are they saying behind

closed doors well here is Jan Talon

future of Life institute's co-founder

and one of the biggest ex risk

billionaires here's the plan for

regulatory interventions his hope for

the foreseeable future is the following

I do think that governments certainly

governments can make things illegal well

you can make Hardware illegal you can

also say that yeah producing graphics

cards above certain capability level is

now illegal and suddenly you have like

much much more Runway as a civilization

do you get into a territory of having to

put surveillance on what code is running

in a data center yeah I mean regulating

software is much much more harder than

Hardware if you like let the more slow

to continue then like the surveillance

has to be more and more pervasive so my

focus for the foreseeable future will be

on kind of regulatory interventions and

kind of like trying to educate lawmakers

and kind of helping and perhaps hiring

lobbyists to try to make the world safer

again keep in mind they have billions of

dollars for this billions so they're not

talking about regulating in the way that

you and I think of regulating they're

talking about global ban on any sort of

training runs take a listen nasty secret

of AI uh field is the AI are not built

they are grown the way you you build the

Frontier Model build the Frontier Model

is you take like two pages of code you

put them in tens of thousands of

graphics cards and let them hum for

months

and then you going have open up the hood

and see like what creature brings out

and what you can can you do with this

creature so it's I think the regulate

like indust the capacity to regulate

things uh and kind of deal with various

liability constraints Etc they apply to

what happens after once once this

creature has been kind of tamed uh and

that's what what the uh fine-tuning and

uh reinforcement learning from Human

feedback Etc is doing and then

productized then how do you deal with

with these issues but uh is is this

where we need the competence of of like

other other Industries but like how can

you avoid the system not escaping during

training run this is this is like a

completely novel issue for this species

and we need need need some other

approaches like just Banning those

training grants the other thing I forgot

to mention is Helen toner said that the

reason everyone backed Sam mman because

remember all the employees signed

letters and tweeted those heart emojis

saying that they wanted to stay at open

AI to stay with Sam Alman even though

apparently he was psychologically

abusing everyone Howen T was saying the

only reason they did that is they were

afraid that open AI would be destroyed

under her leadership or if she did what

she was trying to do and then during the

interview she said well that was not

true it would never have been destroyed

that was not on the table but the thing

is that's not what she was saying while

the whole thing was coming out she was

saying the destruction of the company

could be consistent with the board's

Mission she was trying to either have

anthropic absorb openi or destroy it

which would probably again destroy a lot

of value destroy a lot of these people

their hard work the equity that they

have in the business also if you recall

when open AI researchers were fired for

leaking information out of open AI well

at least one of them had ties to that

effective altruism movement so I do

apologize for that rant I know we've

covered some of the stuff before but

seeing this being discussed other people

in the space other YouTubers talking

about AI I was personally a little bit

concerned that a lot of people seem to

be just taking what she's saying at face

value again I'm more than happy to hear

if she has any proof of anything if she

has any specific things that we can look

at and say okay it does look like this

was correct so far it seems like an

attempt to paint Sam Alman bad light

which again is fine I'm not here to

defend Sam you might not like him you

might prefer that somebody else succeeds

with AI maybe Elon Musk or perhaps

Google or anthropic or I know a lot of

us myself included we believe in open

source we think open source AI is an

important technology obviously comes

with some risks as well but does have a

lot of upside so for example open

sourcing some of the cyber security

technology or at least people getting

together and sharing what they learn

about cyber security helps everyone be

more secure right if every company just

kept what they knew to themselves

everyone would be more exposed more

vulnerable because they aren't sharing

their knowledge with open source

everyone can contribute to the knowledge

potentially helping prevent

vulnerabilities and stuff like that but

it's important to understand that

whatever your views on any of these

companies are that some of the voices

that you're hearing come from this small

minority of people that want everything

shut down that want some sort of a

global surveillance surveillance on gpus

limiting how advanced our chips can be

hitting the pause button indefinitely

and as you've seen even potentially

going after other countries with nuclear

weapons so two nuclear Powers duking it

out to prevent the development of AI I

don't know about you that seems a little

bit crazy doesn't it we know we can

destroy the Earth however many time

dozens of times over with the nuclear

weapons that we have takes a few bad

decisions and everyone's gone but

they're willing to flip the coin on that

to make sure that you know AI doesn't

get developed I don't think this is a

very rational perspective this is very

cultish this is from the blog of italic

butan so he contributed close to a

billion dollars to some of these causes

so he's the person behind ethereum

became very wealthy bought some Dogecoin

or some doggy coin and when it did very

well skyrocketed you know he decided to

you know hey let's throw a billion to

these guys but based on his writing and

he kind of addressed this I don't think

he necessarily is aligned with them he's

not part of that AI Doomer mentality and

here he kind of describes the world as

he sees it and his first image here you

have the anti- technology view right if

we move forward forward the future is

dark scary it's bad technological

progress is bad and safety is behind us

so dystopia ahead safety behind we need

to stop progress we need to unwind the

technology the AI Etc and then there's

the accelerationist view that there's

dangers behind and Utopia ahead so the

idea is that technological progress will

help improve the world Humanity Etc and

if we don't we do have dangers behind

and so I think the EA seemingly adopts

this kind of anti- Technology view or at

least anti- Ai and of course we have the

accelerationists saying you know full

steam ahead let's go in his view he's

describing more as yes there are dangers

behind so we have to progress but we

also have multiple paths forward ahead

some good some bad which I think is very

reasonable so with that said my whole

point in this is don't just approach

this from a stance of is open AI good or

bad is Sam Alman good or bad should we

have ai safety I think most people would

agree AI should be deployed safely

there's a lot of different bad things

that can happen we need to be very

careful how we approach it during the

development of the nuclear bomb someone

suggested there's a chance that the

explosion will trigger a chain reaction

of the entire atmosphere basically

getting set on fire right a fiery death

for the entire world right and that

triggered an investigation into it

people looked into they try to figure

out what what's the chance of that

happening they logically look at that

potential X risk smart people who knew

what they were talking about who had

field expertise who had Education and

Training in that particular field

studied that question and came up with

Solutions that's how it should be we do

the same thing when we roll out any

technology new cars new drugs new

electronic devices new kids toys there's

laws and regulations and how we reduce

those

that's very reasonable no one's here is

against that but if there are people out

there that are convinced that we need to

start nuking other nations out of the

face of the Earth if they're developing

AI they think that we should monitor all

Hardware all gpus all software to ban

training runs to create a global

surveillance system to make sure none of

that happens I hope you agree with me

that maybe just maybe we shouldn't let

those people govern and regulate the

world right that's that's not so crazy I

hope you agree with that said my name is

Wes Rob and thank you for watching