Profile picture
, 3 tweets, 1 min read
I don't think program synthesis will primarily benefit software engineers, at least not at first.

ProgSynth will only be capable of a very small amount of programming ability (near term). Those who stand to benefit from that are those who can't program at all, not the experts.
This is similar to how machine translation benefits most those who don't know the target language at all, but is largely useless to professional translators.

Giving non-programmers the ability to automate routine tasks by "programming" by example will be huge.
This is why Flash Fill in Excel is genius. It's right at the intersection between what the tech can do, how it can produce value, and where the users are
Missing some Tweet in this thread? You can try to force a refresh.

Enjoying this thread?

Keep Current with François Chollet

Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

More from @fchollet see all

Profile picture
François Chollet
@fchollet
There's a big difference between learning to solve problems and learning to look up existing solutions.

When approaching a new problem, it's good to make a serious attempt at solving it yourself from scratch (reinventing the wheel if necessary) before looking up best practices.
First of all, it's practice for problem-solving. It also makes you more familiar with the crux of the problem as well as subtleties you might otherwise be missing -- you will find it much easier to understand other existing solutions after thinking about the problem for a while.
And crucially, approaching problems from first principles means that the act of looking up best practices later will not artificially restrict the space of thoughts you can think.

Because you've already glimpsed the possibilities. You won't merely assume "this is the only way".
Read 3 tweets
Profile picture
François Chollet
@fchollet
In 1993, Vernor Vinge predicted the Singularity would occur before 2030. In 2011, Ray Kurzweil placed it in 2045. In 2016, Masayoshi Son predicted 2047. People nowadays seem more likely to quote 2050.

The common point among these predictions: they're always 30-35 years away.
Why this number? 20 years would be too close and too relatable -- you may have personal 20-years career plans or get-the-kids-through-college plans. 50 years would be too distant and irrelevant -- you'd be very old, perhaps dead.

But 30-35 hits the midpoint between the two.
My own Singularity prediction: I predict that in 2030, futurists will be forecasting the Singularity for 2060-2065.
Read 3 tweets
Profile picture
François Chollet
@fchollet
Imagine an algorithm capable of generating a Wikipedia-like encyclopedia from raw sources (e.g. from the Internet minus Wikipedia), at the same level of factual accuracy, edition quality, scope, and resolution.

When will this be feasible?
Given the recent progress trajectory of NLP and my natural optimism I think something like this is probably 15-20 years away
This poll is about how people perceive NLP progress at this time -- don't see it as a way to actually predict the future. People's perceptions are always wrong
Read 3 tweets

Related threads

Profile picture
Mike Dunford
@questauthority
Good Evening #Threadnought and all the little ships at sea.

New filing:
Reply in support of the motion to dismiss. This, unlike the "Ty Beard: Accidental Notary" motion, is submitted only on behalf of Toye and Rial.

dropbox.com/s/wkzose3zfdo7…
I think we've had 10 filings drop already today, which is over my expectations to the tune of at least 4, and I'm not sure we're done yet. I'll try to unravel things later, and figure out how/what to summarize.

In the meantime, let's take a run through this one.
To refresh: this is the reply to the epic disaster that Ty submitted on Friday. Probably. The Second Amended Complaint complicates this some, to whatever extent that it might differ from Friday's response.

I'm going to talk about that and about bad faith assholery later.
Read 43 tweets
Profile picture
polar
@p01arst0rm
all these tesla fuckboys that spout stuff like "hydrogen combustion/fuel cells are inefficient" "biofuels are a waste or natural land" "diesel fuels are wasteful" really piss me off.

a thread.
now let me get one thing straight. electrical cars are awesome. i think they rock, and the tech has come a long way in a very small amount of time. But if you're stanning tesla as the only solution to climate change, yall wrong
first off, the materials that are mined for use in battery cells have a high cost; in labour, manufacturing, and disposal. theyre also, well, rare earth minerals, meaning there is only so much of them
wired.co.uk/article/lithiu…
Read 26 tweets
Profile picture
Dr Rose Chard
@rrchard1
First-of-its-kind research about Consumers, Vehicles and Energy Integration #CVEI is being discussed today by colleagues from @EnergySysCat . Here's a few of my highlights from the work (1/8)
The conclusions are based on people's actual behaviour not what they say they would do - which for EVs is hard to imagine if you don't have one or have experienced using, charging, paying for one. (2/8)
This is not just w early adopters - “The value of doing trials w mainstream consumers cannot be over-estimated because early-adopters of EVs, or any tech for that matter, are often willing to put up w limitations, faults & costs that no mainstream consumer would ever accept"(3/8)
Read 8 tweets
Profile picture
Dr Sarah Taber
@SarahTaber_bww
IT'S TIME

today's SUGAR FACTS brought to you because the Provosts' gofundme hit the $65K mark @AngProvostFarms @Juneraisincane so now we're in the home stretch!

gofundme.com/provost-farm-l…
There are also some big milestones coming up in the Provosts' court case. It's a great time to catch up on Big Sugar's history of big shadiness!

In some previous SUGAR FACTS we talked a lot the sugar industry's origin story back in slavery times. Today we talk about modern shadiness.

Read 42 tweets
Profile picture
Aaron Hall, 🐍 Professor, NYC, 🇺🇸
@aaronchall
High level #python #tutorial thread:
Python is a high level language that runs bytecode in its virtual machine.
Everything is an object.
It supports OOP up to multiple inheritance and functional programming.
Typing is polymorphic - duck typing is preferred
A Python script is a file with a name like name.py - we indent at four spaces (it matters), comments (after #) don't execute - and is typically written like:

def main(): # function
print("Hello world")

if __name__ == '__main__': # then entry point
main()
A module is a reusable script:

"""this is a docstring at the top of the module - gives help on the module"""

import this # do imports at the top

def main(): # main at top or bottom
"""functions get docstrings too"""

if __name__... always at the bottom, assumed from now on
Read 68 tweets

Trending hashtags

#DeepState #Indians #StrzokPageTexts #IOTA #confidence #DeepShit #RussiaWarCrimes #UraniumOne #CVEI #MARKETProtocol #BUIDL #cryptocurrency #HolyGhost #stackoverflow #ability #EdBuck #PolarVortex #FitIndiaMovement #India #KavanaughStall #CelebrationChurch #NoName #crypto #FusionGPS #KargilWar
Follow Us on Twitter!

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!