No, AI is not going to
destroy humanity.
Jacob Coxon @hilbertspaess
I resigned from Anthropic today. I spent the
last three years doing pretraining research at
both OpenAI and Anthropic. Neither company
is acting responsibly. They are racing straight
to self-improving superintelligence and
gambling with our lives. More thoughts below.
AI NotKillever... and 15K others replied
Evan Hubinger @EvanHub
X.com
Jacob is correct here-we really do earnestly
believe AI could kill all humans! I personally
think it is >10% within the next decade. I believe
Anthropic is trying its best, but we do not yet
have a plan to solve alignment for
superintelligence and are not clearly on track to.
This former anthropic
researcher thinks tha
is going to bring about
the end of the world
in the next few year
And I can't tell if this is a
crazy P.R.
Stunt or if
this is just a
guy who's drunk the Kool
Aid way too much,
because immediately
after Jacob puts this
his former
boss in the
head of alignment and
anthropic pu
puts this up
right here.
And look, I'm no
I'm not an AI
Researcher
but I've been coding with these models since 2019
and I will be the first person to tell yo
it is not intell
And a lot of the TM
anthropics see
think that it
is in
which, in my
is why th
is why the broader world
is starting to hate AI. And
be afraid of it.
However, Jacob's post in
particular reads very suspicious
because he's only really
been an anthropic for
about six weeks,
appear
Now, he'd bee
at Open AI
Before,
but the guy graduated in
2020
And yes,
He's won awar
but he's not an AI.
Researcher
He's not on any
publications you can
He's just no
big of a deal.
And the timing of his
post
along with the Wall Street
all the articles, it just
seems suspiciously like P.R.
Into P.R. Movem
That said, I don't think
Evan's Response is
I think that's just
anthropic drinking their
own Kool Aid a little
bit too much.
One dynamic you should
understand about tech
from a ve
I'm a venture capitalist
saying this is that people
want
and they want to feel
important
and they want to feel successful.
That means th
of a company, you will
try to make your employees feel ve
And you will try
to say that you're working on this
groundbreaking
groundbreaking problem. An
And that's what anthropic
And they've drunk
their
own Kool Aid so much
that people are starting
is going to ta
is going to end
life as we know it.
If you understand how
the arch
the architecture of these
LLM's
the best it will ever be is
the broad level average
of a human. Now,
Could it run
if
if somebody is stupid
and gives an average
human the nuclear
code
Sure. But that's not
Ending the world. That's a person
who is an incompetent leader
ending the world.
I'm not
I'm not saying that there
aren't risks asso
We need
We need to be cognizant
We need to do
We need to do a good job
regulating
We need to actually
understand where we
can and cannot use
But more
we need to understand
the actual technology
that we are deploy
Instead of
Instead of saying it's this hand
magical
magical device that's
gonna solve All of your
problems.
The more complex we make
the more mystiq
the more mystique we put around
it, the more we
the more we try to black box it
and cut off the educatio
and cut off the educational layer as to
what it is an
the more
the more we are going to
have people
the more we are going to
have people thinking
that wil
and a lot of people
drinking thei
Because again, if you do
not understand the actual
if you don't have an
appreciation for the fact
that phi
this is not
this is not intelligence, at
least I don't take a phil
that humans can be
boiled down to
basic numbers,
you
you are not going to have
these problems where
people are going off and
sayin
that A.I. Will end
life as we know it.
And before you come at me and you
you're venture capitalists
you're probably investing in the
You're making money
of A.I.
I'm not. I don't have any
investments in listed A.I. Deal
I think we are on a
massive A.I. Bubble.
This is a 1 to 1 replica for
me of what you see or
saw play out dur
I think it's pretty clear
to say you have a
very early adoption cy
from a consumer side. A product
that hasn't really found
it
It's not producing val
It's not producing value
for the 99% of Americans.
That means
That means that the
the valuations these
companies are trading at are absolutely
ridiculous
I think A.I.
I think A.I. Has a lot of real
value.
we haven't
we haven't really found a
way to comm
way to communicate
that to your everyday
person.
Like, I don't know how my
parents would use
Other than their basic
back and forth chats and
better re
Meanwhile, I
Every day, and I think it's
disrupting how we cod
how we produce things,
and general
how we produce things,
and generally how we build
as a venture capital, as a tech
as a venture capital, as a tech
I think it is a transformative technology.
I just don't th
I just don't think that that correlate
and I certainly don't
and I certainly don't think
that it is going to go
Rogan in the wo
when A.I. Goes
For me personally,
For me personally, what
that usually looks like is I
am contr
And A.I.
If you
If you use a tool poorly, it
will go wr
it will go wrong. That is how
tools work.
But just in the same way
we can use A.I. Poorly,
we can use general
There are
There are hundreds of
thousands of stories of
people be
people being breached, their data
massive security th
massive security threats
due to poo
and poor deployments of the
A.I. Is not goi
A.I. Is not going to be any
different th
But no, I
But no, it is not going to
be the end of all huma
And All of human