Alexandr Wang
@alexandr_wang
Followers
143,951
Following
704
Media
198
Statuses
1,710
ceo at @scale_ai . rational in the fullness of time
San Francisco, CA
Joined June 2012
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Mother's Day
• 1167929 Tweets
Arsenal
• 355563 Tweets
Feliz Dia
• 309445 Tweets
Mães
• 264314 Tweets
Moms
• 259220 Tweets
#鬼滅の刃
• 173388 Tweets
Manchester United
• 110617 Tweets
NAYEON
• 90107 Tweets
#MUNARS
• 69554 Tweets
Saka
• 47175 Tweets
Casemiro
• 45796 Tweets
Karagümrük
• 41651 Tweets
Old Trafford
• 37759 Tweets
Man U
• 32490 Tweets
ヒロアカ
• 32366 Tweets
Amad
• 30586 Tweets
Garnacho
• 28526 Tweets
Evans
• 26647 Tweets
Arapça
• 25556 Tweets
Antony
• 23445 Tweets
Arteta
• 22643 Tweets
Trossard
• 22530 Tweets
Hojlund
• 18909 Tweets
ارسنال
• 15321 Tweets
Onana
• 14686 Tweets
Amrabat
• 14104 Tweets
Partey
• 13846 Tweets
アニオリ
• 11762 Tweets
伊黒さん
• 11084 Tweets
The greatest modern tragedy is:
The most brilliant and determined people often do not work on the world’s most urgent problems.
We are on the brink of geopolitical crisis, and yet the smartest people I know continue to work on crypto, finance, vertical SaaS, etc.
787
1K
10K
Life hack: rewire yourself to enjoy grinding.
MIT did this for me & most classmates. The school embrace the grind because there is no other way.
The harder you work early in your career, the faster you’ll compound. The payoff later in life can be astronomical (>10x).
236
653
6K
PREDICTION: Google’s $15B deal with Apple to be the default search on iPhone will be re-negotiated and be a bidding war between MSFT/Bing and Google.
It will become at least $25B, if not more.
If MSFT is willing to spend $10B on OpenAI, they’ll spend even more here.
187
341
5K
i was doing math problems on the plane (usamo 2023!) and the flight attendant asked me if I was doing schoolwork (flattering age-wise she thought I was in school)
it’s a total shame that most people stop doing any technical or quantitative deep work post-school, esp on pen&paper
138
230
4K
the next 2-3 years of AI are definitively going to define the coming 2-3 decades of the world
for those in technology: you live a lifetime for a moment like this—don’t waste it; don’t be lazy
there are decades where nothing happens, and weeks where decades happen.
83
690
4K
the most valuable skill in the world is systems engineering:
the ability to debug, understand, and improve a complex system with limited/poor measurement
THIS is what makes great scientists, engineers, PMs, operators, doctors & investors
not truly taught in school outside STEM
115
447
3K
more has happened in the last year of AI than the prior 10
we are unmistakably in the fiery takeoff of the most important technology of the rest of our lives
everybody—governments, citizens, technologists—is awaiting w/bated breath (some helplessly) the next version of humanity
118
421
3K
.
@elonmusk
just came onstage with Dave Chappelle at Chase Center
Elon was booed continuously for 5 minutes
Finally, he asks Dave Chappelle to get on twitter
SF is most definitely back!
276
125
3K
I’m not a parent nor a child development expert…
But it does seem like the best way to avoid having Normie kids is to allow them to play video games
…and it’s definitely the best way to get them interested in computers
Why do parents crusade against video games so much?
532
154
3K
biggest red flag on a résumé is a spotless track record of high brand companies
it indicates:
- not sticking thru adversity
- risk adverseness
these people are generally actors—they “play the part” of a builder without actually being one
warts are a sign of realness
172
183
3K
obsession goes a very long way, much further than intelligence does.
64
457
3K
my interests roughly by age
0-12—math+chess all the time
13-17—math+algos+physics all the time
17-25—stopped doing math. tried to be cool. coded, read literature, played poker, traveled, went to concerts/parties
25+—math all the time
as I tell my nephew, math will set you free
87
151
3K
College is closer to a 4-year party than a career-preparation program.
Most curriculum is totally unrelated to your future job—most jobs rely on skills that are entirely absent from college (planning, leading, etc).
But, college is a ton of fun. Just know why you’re going.
208
202
2K
The greatest luxury in life is being in an environment where your only responsibility is to learn
This might sound like college, but college is not a place to learn most useful things
The sweet spot is apprenticeship in a dynamic environment—a hypergrowth startup for example
94
270
2K
working hard is infectious
being around people who are always working makes it shockingly easy to work really hard
inversely, surrounding yourself with leisurely or lazy people will make it nearly impossible to work hard
choose your own adventure—but be intentional
45
268
2K
there are so many engineers (literally 10s of thousands) at big tech companies who are working on things that fundamentally don't matter and nobody cares about
this is because at "the top", there's politics where it's better to be nice vs right, and deep inefficiency persists
113
174
2K
If you want to be a historian, go into finance.
If you want to write the future, go to a startup.
Financiers study the ideas that already exist and pick the profitable ones. They study how things have been done already, and cap their imagination.
Startups invent the future.
67
540
2K
Advice to my younger self:
- surround yourself with truly excellent people. top 1% in some important way
- find something to work on for a LONG time (10+ yrs). jumping around doesn’t beget mastery
- work harder than you think you should. hard work is a habit best learned young
65
261
2K
i have been weird my whole life
everybody i’ve met who I respect has been weird
the coolest people are usually also the weirdest
weird is one of the best words in the english dictionary
143
230
2K
immigrants have some of the hardest lives
immigrants are often discriminated against, have to work dramatically harder, are culturally alienated, and 1000s of miles away from everyone they know
and yet come for opportunity and promise of a better life
should be celebrated more
97
230
2K
Unintuitive fact of engineering:
Big teams (>8 people) ship less than small teams. Not only less per capita—less overall!
A big team that is not a fractal of small teams will lose. David vs Goliath is real.
Managers and investors—understand this before asking to grow teams.
91
232
2K
one thing that most AI hypers don’t appreciate is just how slowly most of the world changes
most workloads at big banks still are 80%+ on-prem, despite the cloud existing for 16+ years
true AI-everywhere adoption will take decades, despite how mind-blowing the tech is
172
199
2K
Much of Silicon Valley is the reshuffling of uninspired, unremarkable individuals with one Great Brand on their resume, being recycled through tens of other Aspiring Great Brands.
Do not outsource your talent evaluation to Brand identification, otherwise your company will suck.
57
142
2K
Last week,
@scale_AI
hosted a Generative AI hackathon.
It was the purest expression of builder energy, and an omen that the dark days of tech stagnation are over.
Roughly ~300 hackers converged @ our office in SF, and a day full of frenzy ensued
🧵 of cool things that happened
35
228
2K
urgency is a hell of a drug
if you find yourself addicted to the feeling that every moment counts—cherish it.
urgency will feel quite unreasonable to most. as a general rule, most of the world is slow.
ignore them. run fast. do something that shoots electricity into your veins
50
260
2K
The first thing you have to unlearn as an Adult is:
Stop worrying about the BS everyone else is worried about.
College apps & similar games train you to think the crowds are right and we should all care about the same thing
Not true. You win by focusing on new & secret metrics
43
201
2K
The real problem with other cities (LA, NYC, Miami) relative to SF is that people are severely over-socialized.
You need time alone to think, reflect & build plans to do anything important.
In LA/NYC/Miami, the constant social rollercoaster steamrolls all individual thought.
245
122
2K
one of the biggest lies of psychology is that introversion vs extroversion are innate traits
extroversion can be learned—given a sufficient number of positive social interactions, you can become an extrovert
and introverts who become extroverts are generally unstoppable ;)
205
151
2K
awkward truth:
engineering productivity is far more correlated with culture than team size
there are startups w/ 10 engineers that ship more for their customers than big co engineering teams of 1000 (ex: Midjourney)
you cannot throw people and/or money at building great things
93
246
2K
truly, bias for action—more than any other trait—is the trait that seems most correlated with success and impact
it is SO much better to be constantly shipping, making mistakes, and learning than to be smart, cynical, paralyzed & inert
get out there—go do some shit!
22
233
2K
One of my favorite interview questions is: "What’s the hardest you’ve ever worked on something?"
It's an extremely efficient filter for the kinds of people I like to work with.
What are your favorite interview questions?
115
114
2K
the 2 years of life where the greatest human potential is lost are the first 2 years entering the workforce
everyone has a sharp realization that nobody actually cares about their work—the workplace is a theater to avoid responsibility & pass blame
startups are different though
66
123
2K
hope I’m wrong, but my estimate is there is ~10% chance this is the start of WW3
chaos is a ladder & the start of every major war has been a series of unlikely, unconnected tail events
the US is key to major fronts of the evolving war—American leadership has never mattered more
228
82
2K
the fact that Google market cap fell $100B in a day after being *slightly* late to RLHF-enhanced LLMs (ChatGPT-like models)—after inventing the Transformer—is a great lesson that you cannot take your eye off the ball.
you must have stellar execution at all times—or it’ll hurt
76
124
1K
Bit of startup advice which I rarely hear:
Spend the money on a nice office.
Be frugal on just about everything else, but get a nice office.
It's hard to appreciate just how much your office becomes a representation of your company. Plus, you're going to spend 10k+ hrs in it.
42
153
1K
funny misalignment:
- most employees (esp high performers) want SMALLER teams—fewer people to deal with, more work done
- most managers want BIGGER teams—more people = more power, more to “manage”
you can tell how strong a manager is by how often they ask for more headcount
57
130
1K
only a very small % of the population truly understands how quickly AI will improve in the coming years
and how much will change as AI models solve reasoning, coding, multimodality, tools, etc
most predictions are wrong. even with all the hype, AI is still underrated
93
180
1K
Let’s not lie to ourselves about American Dynamism
The fact that
@a16z
doesn’t include any AI, chip, or cyber companies, despite these being the top 3 tech competitions w/China, is laughable
We won’t beat China with vertical SaaS
It’s just them schilling their portfolio
82
81
1K
Large language models are magical. But using them in production has been tricky, until now.
I’m excited to share ✨Spellbook🪄✨— the platform for LLM apps from
@scale_AI
. 🧵
49
132
1K
every great startup I know started as a nucleus of hardworking, insecure nerds
55
93
1K
How to hire people better than you:
1—Clear most of your calendar
2—List the most talented people you know
3—Meet up with each one
4—Paint a genuine vision and progress against it
5—Explain their potential impact on the vision
6—Listen & be humble
7—Repeat 3-6 every 2 months
14
171
1K
let’s all be honest—working during COVID sucked. a lot.
get your ass back in an office and rediscover what it feels like to love your work, and work with a team you love
not too many other life hacks out there that are this easy
2K
70
1K
The cultural prohibition of grinding & hard work will be the road to hell for America.
If you love America, you’d realize our future is contingent on industrialism, not hedonism.
83
116
1K
One of the saddest societal norms is losing curiosity at age 22.
We are encouraged to in explore through the end of college, but then fiercely pivot towards exploitation—“go get a career!”
People intellectually complexity until 22, but then we simplify.
It’s a tragedy.
85
129
1K
I’m seeing many fall into the “self-driving trap” w/Gen AI
The self-driving trap is seeing shiny demos & thinking said demos will reach 100% reliability for prod w/in a few years
Many proposed Gen AI use cases need 100% reliability, and thinking that’ll come soon is a mistake
83
146
1K
despite all the hate & bs on twitter, san francisco is still the heart of AI, the biggest technological shift since the dawn of the internet
the tech people who loudly moved out of sf in 2020 have ended up missing the most important thing to happen in tech in decades
oh well 🙃
96
61
1K
My biggest mistakes in life (so far):
1—Forgetting who is truly there for me (the Real ones)
2—Not telling them I appreciate & love them
3—Getting off the learning treadmill at times. NEVER stop voraciously learning
4—Not asking for help when I’ve needed it
5—Not writing enough
38
103
1K
one of the secrets you learn from those older and wiser than you:
speak less. say more when you speak (ie pack more information into fewer words).
it’s easy in youth to think speaking is winning. it isn’t—in fact it’s usually losing.
43
136
1K
Peter Thiel’s criticism was famously “we asked for flying cars and got 140 characters”
Well now, we have an on-demand artist, on-demand expert chatbot, on-demand copywriter, on-demand pair programmer…
We’re entering the sci-fi age with AI. Buckle up!
63
92
1K
I’m always shocked by smart, young, driven people who choose a company because of “mentorship”.
Yes, it is good to have some guidance, but most driven folks succeed through figuring sh*t out on their own.
Being around smart young people >>> mentors, even for energy alone.
108
112
1K
The way that Bezos and the AMZN team derived AWS was by asking:
“If compute costs will trend to 0 in perpetuity, what business would you build?”
The purity is staggering. Today’s AI parallel is:
“If LLMs will continue to gain more capability, what business would you build?”
28
85
1K
the biggest mistake big companies make is thinking most problems can be solved by money
in reality, the best problems can only be solved by blood, sweat, tears, spirit & an overwhelming sense of purpose
“we think about this a lot because it’s not a job, it’s our life”—steve…
37
106
1K
top recommendations for 2023:
pour your heart & soul into your work and passion
live without regret, don’t let lack of effort be an excuse
go for depth rather than breadth in your relationships. invest in your real ones
fight for what you believe in
love more than you hate
17
139
1K
everything i've learned about operating:
trust your gut. act quickly. don't let problems stew.
24
84
1K
Chatted with a European founder today:
“You’re so lucky to live in SF. You’re surrounded by brilliant, friendly, insightful people all trying to do crazy things. You’re surrounded by this light that stimulates your brain.”
Perspective is powerful. Feeling lucky to live in SF 🥰
38
36
1K
I turned 26 this week🥳
Every year, I more deeply appreciate wisdom & mastery
Youth is energy & a willingness to brute force anything. With wisdom, you learn to be one with nature—push smarter
The ideal is a mix of both—the energy of a young bull w/the knowing eye of a master
52
46
1K
perhaps the most underrated & important skill:
being able to evaluate people on their intrinsics (i.e. their intellect, talent, attitude—their true self) VS their extrinsics (i.e. resumé, brands, external signals)
i've met maybe 5 people in my whole life who are great at this
55
104
1K
something
@paulg
once told me (paraphrased):
“one of the reasons I think YC worked is because I truly do believe in everyone and got them to believe in themselves.
and maybe I’m a fool for believing in everybody, but I really do!”
PG has the most supportive & brilliant warmth
20
59
1K
Not to call out all the web3 folks, but there’s a stark contrast between web3 and AI.
web3 at peak: “all the smartest people are working on it, cool stuff will happen”
AI at peak (today): “you can literally perform magic, oh and also all the smartest people are working on it”
49
72
1K
The reason to work at a successful startup before starting your own is to have a really good initialization.
Most startups are broken in many ways, but are incredibly effective at the most significant bits.
It pays to be imprinted with a winning culture.
15
122
1K
The scarcest human resource is full-hearted commitment.
What could you accomplish if you committed headfirst, w/no exit strategy, into a critical mission for the next 10 years?
The answer is exponentially more than if you’re half-in-half-out.
Dreams die in the zone of “maybe”.
31
131
1K
one thing you learn as you get older:
fewer deep relationships >>>> many shallow relationships
people who do truly meaningful work do so in small squads of deep contact
those with many shallow relationships end up becoming brokers in one way or another
24
95
1K
I have become mildly obsessed with the process of learning new skills.
In many ways, it feels like one of the few fulfilling & sustainable meta-activities.
What’s the most rewarding skill you’ve ever learned?
219
43
985
the first—and still only—killer use case of ChatGPT is doing homework
nearly every student is using it everyday
but I haven’t heard of anyone else using it consistently
209
35
982
after traveling like crazy for the past few years
i’ve realized most of the people who are fun to talk to are in the Bay Area
which makes living in SF extremely convenient :)
64
20
983
I truly wake up every morning feeling like we are in one of the most important moments of human history.
70 years of AI summers and winters have culminated in this moment, in which we will see the greatest shift of how people interact with technology since electricity.
game on!
36
82
957
there’s more than enough people starting companies
there’s not enough people finishing them
the salient milestone needs to be building something meaningful that people love
not putting “founder” in your twitter bio
36
83
953
Attention is all you really have.
Be careful how you allocate it. Social media is built to hijack and fray your attention.
Give your attention to long term goals and long term people. All else is noise.
[My sequel to “Attention is All You Need”, the seminal Transformer paper]
23
139
949
For those who say AI isn’t creative as it can’t have original thought—
Humans are no better. Creativity is the act of mimetic recombination. We combine the set of ideas we’re exposed to (“influences”) to form new ones
And like AI, humans often don’t realize when they’re copying
86
112
945
one reason why math can be better for your brain than coding:
coding is constrained in instantiating the imminently possible, thereby limiting the scale & abstraction of thought
math forces you to think about the limits of human thought (ex: cardinalities of infinity)
61
93
942
I wanted to preview of one of the coolest products from the
@scale_AI
labs.
We're enabling marketers to AI-generate UNLIMITED and INFINITELY CREATIVE images of their products for:
- ad creatives
- brand campaigns
- social media
Every image in this thread is AI generated🧵
36
105
911
Want to fine-tune Llama2 for your use case on your data?
ICYMI—
@scale_ai
open-sourced LLM Engine, the easiest way to fine-tune and serve Llama2 and other open-source models
See 🧵👇
15
126
903
A hill I will die on:
Tea is significantly better than coffee
You have a GRADIENT of caffeine options—zero (herbal), low (hojicha), mid (green), high (black, earl gray)
The taste profiles also are far more diverse, subtle, and interesting—rather than assaulting w coffee & milk
156
54
898
kinda surreal
but this is very much a team honor for one of the best teams in ai (✨the
@scale_AI
team✨)
and the incredible innovation of our customers, who inspire us every day
and the journey is nowhere close to over :) here’s to the next leg!
67
45
885
after 5 years of hard work in any job, you’ll accumulate 10,000 hours of practice, and be an expert/master
the fact that most people leave gigs before that point is self-sabotaging—it’s only after 5 years you start getting good
most people’s attention span is simply too short
84
96
888
it’s easy to forget when you’re young that life is a marathon
43
75
876
weird phenomenon—
when you're young, you say what you think—you don't know any better!
then as you get older, you start to say not what you think, but what others want you to say.
eventually, you learn that true power IS saying exactly what you think, and revert to childhood.
60
86
865
The one thing we know works for education is 1-1 tutoring vs one-to-many classroom
1-1 tutoring with humans for everyone is economically infeasible
But the promise of AI and ChatGPT systems will be 1-1 tutoring for everyone in the world
Future is bright :)
54
134
877
Waking up early (pre 6am) & starting the day with a walk has been one of the best hacks to increase my quality of thought
There’s research that shows activating your visual cortex w/a walk turns on your brain, & my thoughts in the morning are clearer, calmer, and more deliberate
35
56
870
we're starting to see top companies spend the same amount on RLHF and compute in training ChatGPT-like LLMs
for example, OpenAI hired >1000 devs to RLHF their code models
crazy—but soon companies will start spending $ hundreds of Ms or $ billions on RLHF, just as w/compute
37
105
880
"We used to move faster before the pandemic,"
I've heard this 10+ over the past few weeks.
Why?
The lethargy of remote work is palpable. There's no electricity, spirit, comradery, and magic.
This will be the story of winners and losers in 2023—velocity always wins.
75
70
871
We’re currently watching the previously impervious-seeming tech giants being meaningfully threatened, moats be damned:
- Meta by TikTok
- Google by ChatGPT
The march of technology and consumer habits shows no mercy.
The only true moat is velocity and customer obsession.
45
77
857
We are at a critical turning point for humanity:
Children born today are likely to have more AI friends than human friends
AI friends are going to be more reliable, conciliatory, and considerate
What does this means for childhood development and social norms? We will find out
138
120
802
The economic opportunity of using LLMs/generative AI to build a next-gen search engine might be bigger than the opportunity for using LLMs for literally everything else, combined
The real Q is—will Google move fast enough to win? Or will someone else (Bing/OpenAI, startup)?
66
81
815
it is crazy to the degree to which mindset & attitude is the key indicator for how much impact you’ll have at a startup—WAY more than raw talent.
if you roll up your sleeves, have nothing to prove, & are optimistic even in the face of daunting facts…
you will move mountains
27
95
802
I'm publishing a call to action: The AI War and How to Win It.
AI for national security will define the future of our world. Either the USA wins, or our authoritarian adversaries do.
I walk through The AI War, The China Threat, and How to Win It.
🧵
47
125
769
.
@karpathy
just published a wonderful video as an introduction to LLMs.
Go watch it if you’re bored on thanksgiving!
ICYMI, he used ChatGPT to definitively determine
@scale_AI
is worth $150B today and $2T in 2 years
The AI does not lie 🤷
New YouTube video: 1hr general-audience introduction to Large Language Models
Based on a 30min talk I gave recently; It tries to be non-technical intro, covers mental models for LLM inference, training, finetuning, the emerging LLM OS and LLM Security.
585
3K
18K
30
44
756
Last night while working late, I remembered this striking quote from the Steve Jobs biography:
“I sent emails… at 2am and batted things around. We think about this a lot because it’s not a job, it’s our life.”
When I read this sentence, it struck a deep chord. It’s our life.
19
73
746
AI is too hyped right now.
The technology is will fundamentally change humanity for the better.
But the fever-pitch-excitement and every-VC-pivoting-to-AI will invariably lead to disappointment and a deflated hype cycle.
Let’s focus on the long game and real impact.
41
51
747
We’re entering a new era
1980s: Mass Media (MM)📺
2000s: User-Generated Content (UGC)📱
2020s: AI Generated Content (AIGC)🤖
MM—>UGC was an explosion in content & virality emerged
UGC—>AIGC will be a similar explosion & *individualization* to audiences of one will emerge
37
119
749
The nerds vs the cool kids, a short thread 🧵
The nerds are jealous of the cool kids for being, well, cool.
The cool kids are jealous of the nerds for the ability to build.
1/n
34
110
753
most great operators I know are inherently anxious
only the paranoid survive
generally speaking, not well appreciated enough how valuable pathologized psychological traits can be
46
50
749
Something I wish existed:
Immediately in a conversation with someone, I could know the set of topics that we’re both deeply curious and/or knowledge about.
Conferences are the weakest form of this—ostensibly everyone is interested in the conf topic.
Need the generic solution!
109
24
733
Quote from a friend visiting SF this week:
“I learned more in 2 days in SF than in 2 months in NYC”
You tell me where you should be based 🙃
66
43
736
