The academic peer review system is not going to survive this without changes.

Fun fact: A lot of "AI" in industry is often a very simple algorithm designed to deliver "good enough" results at scale. When will we start seeing real AI in products? How many companies will figure out how to justify the cost at scale?

The real magic in AI is not achieving human-like performance with massive compute budgets -- it's taking neural networks with billions of parameters and shrinking them down to 1/20th the size so they can actually be useful.

What happens next? Identity theft over Zoom calls in a digital and remote work world, that's what. Give it a year or two. https://t.co/PTKn95IfoQ

If a small team created neural networks that are 10x-20x more efficient than state of the art models, we'd never hear about it with all the benchmark leaderboards and hype being dominated by models that use 100x the compute budget accessible to smaller research teams. #science

Does listening to thousands of hours of music somehow make you a musician? Does reading Wikipedia make you an author? Does watching YouTube make you a video editor?

If artificial neural networks are supposed to emulate the brain, and every day more research is finding parallels, why do we design training tasks that would fail to teach a human and expect them to work?

Brokers artificially limiting retail investors while big money gets to play by their own rules *in an international movement* is a wakeup call for everyone. This is not, and has never been, a fair and legal market. #GMEstock

Most people in the ML industry right now: Replace X model with Transformer, publish new "SotA" paper on language/vision/whatever. Very few people in ML industry: "How do Transformers represent data? Can we do better?" (Yes!)

It's going to be really funny when the general deep-learning community finally realizes that all the "hacks" they rely on to train DL models are the very reason they don't generalize on unseen data. Fits the data, not the world.

We're thrilled to announce that Axonify made the 2020 list of Best Workplaces in Technology! This is our second nod from the organization this year. We also made the 2020 list of Best Workplaces for Today's Youth.

There is something exhilarating in failing and being able to tell exactly what went wrong. It feels more empowering than succeeding and not knowing why.

You understand a thing if you can recreate it -- not in the weak sense of being able to *copy* it, but in the strong sense of being able to generate deliberate variations of it. Including improvements. You understand a thing if you can recreate it *better*.

Toronto Star: 'We're hustlers': Amid coronavirus fears, this couple has made more than $100000 reselling Lysol wipes. https://t.co/ZoWX4U1w6g

As one of the fastest spreading pandemics continues to ravage the world, there's a special group of people looking to profit from the fear and desperation. Are we really sure that we want AI that behaves like humans? It can do better.

The parallels between the research community Hive Mind for artificial general intelligence and Twitch Plays Pokemon is surreal. https://t.co/QSRdrTS4vJ I look forward to when the Hive Mind eventually discovers that we've had all the pieces to do it since 2013.

I keep reading about how algorithms can be, or are, biased. Does that mean people also care about humans being incredibly biased and prejudiced against certain groups? Can we please try to fix that?

Great news! We just won a 2019 ARTIFICIAL INTELLIGENCE BREAKTHROUGH AWARD for "Best AI-based Solution for Retail". Axonify believes in AI with purpose. And we’ve been doing it since day one. #AI #Retail https://t.co/AudGuHzb5F

Our CEO @CarolLeaman shares her perspective on what it’s like to lead people, embracing our power as women, and what we need to get better at in leadership. Listen to the latest episode of the Inner Circle Podcast with @CarrieADoll #cdntech https://t.co/yX39tprGsH

My two-year-old said "I'm a cookie, eat me!" I stood confused, so she said "you're a cookie!" and bit me. Hmm.