Profile picture
François Chollet
@fchollet
, 4 tweets, 1 min read Read on Twitter
I've been following the work of Meredith (@mer__edith) for several years now. She, alongside others such as Kate (@katecrawford), has been a real pioneer & thought leader on AI ethics, especially when it comes to ML bias, diversity, & inclusion, topics that I'm following closely.
I believe that her work is critical to Google's mission, as an AI-first company that is highly reliant on being fully trusted by a diverse user base of billions of people. Meredith's work is a net benefit for Google's users, for Google's products, and for Google as a company.
I hope that she will be able to carry on her important work on the social impact & ethics of ML. Thank you for your valuable contributions so far -- here's hoping that this is just the beginning.
As everything else I tweet, these comments are made in a personal capacity.
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to François Chollet
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content 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
This is how you implement a network in Chainer. Chainer, the original eager-first deep learning framework, has had this API since launch, in mid-2015.

When PyTorch got started, it followed the Chainer template (in fact, the prototype of PyTorch was literally a fork of Chainer).
Nearly every day, I am getting ignorant messages saying, "PyTorch is an original innovation that TensorFlow/Keras copied". This is incorrect. Subclassing is a fairly obvious way to do things in Python, and Chainer had this API first. Many others followed.
I had been looking at adding a Model subclassing API to Keras as soon as late 2015 (before the Functional API even existed, and over a year before being aware of PyTorch), inspired by Chainer. Our first discussions about adding an eager execution mode also predate PyTorch.
Read 6 tweets
Profile picture
François Chollet
@fchollet
When you're using tf.keras in TensorFlow 2.0, training your models with `fit` is one option, but you can also write a custom training loop. It takes about 10 lines.
If your model creates additional losses during its forward pass (i.e. if you called `self.add_loss` in a custom layer), collect them in one line like this.
The advantage of using `fit` is that it packs a lot of functionality in a convenient interface: a progress bar, easy reporting of metrics such as precision & recall, easy management of class weights for imbalanced classification, callbacks for model checkpointing and more...
Read 3 tweets
Profile picture
François Chollet
@fchollet
Just like everything else, thinking takes practice.
Thinking is recombining largely preexisting concepts following largely preexisting patterns. Infrequently do we generate brand new ones -- it takes a lot of effort. As such, deep thinking requires practice in order to accumulate a dense network of these concepts and patterns.
Of course, it also requires to read books. It would be challenging to have to recreate thousands of years worth of accumulated thought building blocks on your own.
Read 4 tweets

Related threads

Profile picture
Maria Luciana Axente #AIforGood
@maria_axente
Dear #AI ethics community - can you point me towards some interesting yet not enough in limelight real life applications of ethical principles for AI and AS with their ACTUAL impact - both good or bad?
Many thanks
ping @DorotheaBaur @EmergTechEthics @David_Gunkel @IEthics
Cc @vdignum @RosariaTaddeo @JoshCowls @boom_lawyered @RDBinns @Reema__Patel
Also cc @baxterkb @MiaD @frossi_t @j2bryson @jovialjoy @lilianedwards
Read 3 tweets
Profile picture
Chad Loder ❁
@chadloder
Let's talk about the dangers of #AI and #MachineLearning hype. We over-hype AI such that business owners, and policy makers have unrealistic faith in them.

As a result, the public becomes unwitting beta testers for dangerous and poorly-thought-out AI applications.
#AI has become a hammer in search of nails.

"Can #AI detect emotions better than humans?" - the answer is no.

AI still has a hard time figuring out whether dark-skinned faces are, in fact, faces. But we're deploying this stuff, because, capitalism.

datasciencecentral.com/profiles/blogs…
The same #AI hype pattern is repeated endlessly.

1. Company makes hyped claims about its AI. None of it works.
2. Human workers are thrown into the gears behind the scenes.
3. Company hopes its AI will catch up with its marketing.
4. Humans get hurt.

techcrunch.com/2019/01/10/ama…
Read 6 tweets
Profile picture
Chad Loder ❁
@chadloder
In the same way that Silicon Valley has tried to take a reductionist, metrics-driven approach to #diversity, we're seeing an attempt to reduce discussions about bias in #AI to a purely mathematical, "garbage in/garbage out" problem , e.g. fairness in training data.
We can't talk about AI #ethics in isolation from social context, in isolation from racism, from policing, from colonialism.

We have to talk about WHO is creating these AI systems

WHY they are being created

WHO runs them

for WHOSE benefit

and against WHOM are they being used
If we talk about bias in #AI and #machinelearning from the framing that "all conceivable technologies and business models MUST exist and WILL exist", then we're reduced to talking about harm reduction in the tiniest margins at the edge of giant machines of #SurveillanceCapitalism
Read 5 tweets
Profile picture
🇺🇸 TX Intense TV 🇺🇸
@TxIntenseRadio
@realDonaldTrump signed an executive order on 2/11/19 directing federal agencies to prioritize research & development in the field of AI

The American AI Initiative, is intended to enhance national & economic security & improve Americans' quality of life.

foxnews.com/tech/trump-to-…
President @realDonaldTrump laid out 6 objectives in the Executive Order on Maintaining American Leadership in AI for 6 Departments/Agencies

Secretaries of:
- Defense
- Commerce
- Health & Human Services
- Energy
&
- the Administrator of NASA
- the National Science Foundation
#AI
Executive Order on Maintaining American Leadership in AI
(d)The US must foster public trust & confidence in AI technologies & protect civil liberties, privacy & American values in their application in order to fully realize the potential of AI technologies for the American people
Read 44 tweets
Profile picture
Duncan Ross
@duncan3ross
So you're a believer in #AI #Ethics? We all are? How come there are so many twitter arguments about it then? The answer (in a Myers/Briggs* style) is because we have different attitudes to certain key questions. Thread. 1/15
Our first disagreement is around People and process (P) vs Technology (T). Do you focus on using frameworks, or on better tech to resolve the issue? 2/15
There are great frameworks out there: @AdaLovelaceInst (turing.ac.uk/research/data-…) and @Floridi, work we've done at @DataKindUK (datakind.org/blog/doing-dat…) with @christinelhenry, the EU (ec.europa.eu/digital-single…) and more 3/15
Read 15 tweets
Profile picture
Janosch Delcker
@JanoschDelcker
THREAD: There’s lots of talks about artificial intelligence and ethics these days — how to come up with principles for the use of #AI, whether there should be binding laws regulating it or whether broader guidelines could be enough etc pp.
2/ I’d like to point you to an — overlooked, as often — academic paper that came out earlier this month. There’s lots of op-eds and papers written on the topic. No one has the time to read them all (and to be frank, many aren’t worth it.) This one’s different:
3/ The paper’s titled “Constitutional democracy and technology in the age of artificial intelligence” and it’s written by @PaulNemitz , a high-ranking European Commission official who was one of the architects behind Europe’s sweeping privacy law, the GDPR papers.ssrn.com/sol3/papers.cf…
Read 15 tweets

Trending hashtags

#NoseBleed #Quechuan #Chatbot #CRISPR #patientsafety #MothersDay2019 #4IR #pregnancy #mammogram #CircularEconomy #Trolls #neurology #AI #blockchain #Breathlessness #ArtificialIntelligence #NAFTA #TheBeginningIsNear #Bears #Cyclotron #ProjectMaven #ArtificialInteligence #MRI #Gaza #computer
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!