There are many machine learning developers in the world. Only one person has been mentored by both Andrew and Yoshua, invented a new kind of artificial neural network, contributed to or led research at Apple, and still has less than 40 candles to blow out on their birthday cake.
He will be working for the DeepMind research team after Apple let him walk out the door.
Ian Goodfellow is his name. Allowing him to walk away from your California-based, Silicon Valley elite, ultra-progressive technology company because he disagrees with your in-office work demands is a scenario that is so dumb I can't believe GPT-3 didn't do it.
You can get a weekly recap of our favorite stories.
It's like letting Tom Brady or Michael Jordan leave your team over a disagreement with the team owner.
Goodfellow and his team can work wherever they please. If they think they can code a better machine learning model from the International Space Station, you should probably build a rocket.
Put things into perspective. I can sense that our C-suite readers are going to prepare a good speech in response to this opinion piece.
I don't believe anyone deserves special treatment at Apple. For that matter, any other company.
Ian Goodfellow's contributions to the field of machine learning cannot be overstated. Office hours are a pretty silly thing to lose any talented developer over, and it's even more ridiculous to let your director walk because you think in-person smiles are important.
Losing Goodfellow is a huge blow to Apple.
Goodfellow's greatest claim to scientific fame is that he was a part of the team that invented the GAN.
A neural network learns how to create content by fooling itself and eventually humans.
If you hear about an artificial intelligence that can generate text, write poetry, create images, or produce its own original music, you're almost certainly hearing about a GAN.
GANs work by pitting two neural networks against each other. Without GANs, humans would have to fine-tune every iteration like sanding with fine paper.
One network is creating and another is discriminating. The second network is a bouncer that blocks a lot of useless output before it even has a chance to manifest.
He continued to work on computer vision and security issues after Goodfellow et al. invented the GAN. He went on a tour at OpenAI, a think tank that is funded by Microsoft and is where some of the world's best minds try to figure out how to invent and control a human-level artificial intelligence.
There is only one other company on the planet that is as invested in AGI and full of well-known talent as DeepMind.
Many of us in tech journalism were surprised when Ian Goodfellow became the director of machine learning at Apple. It made sense for Goodfellow, since it seemed like a well-deserved promotion, and it didn't seem like something that would happen.
Today, as Goodfellow is re-welcomed to the fold, it looks like the smartest player on the board.
DeepMind is a version of Openai. DeepMind is more focused on creating an artificial intelligence that can do anything a human can do without needing to be trained over and over again.
This could bite Apple in the butt.
Five years ago, if you wanted to make a joke about a virtual assistant, you had to say "Siri". All of my readers were familiar with that artificial intelligence.
I'm better off using an amplification device for name recognition. Nobody has forgotten about Siri. At least not yet.
DeepMind has a new system called GATO. I don't think it's on the path to AGI with GATO, but that's a debate for a different article.
If DeepMind can overcome the problem of giant-scale models and bias, GATO could be very popular.
A version of Siri that could do a thousand different tasks on your behalf. Virtual assistants are doing web searches and open apps for us. It might feel like you can do a lot of things, but telling you what time it is, how many messages you have, and what the capital of Nebraska is is the same task.
I'm talking about a version of Siri that could control a robot capable of washing your dishes, while simultaneously identifying high-weed areas in your front lawn, while also generating a completely original cartoon for your kids to watch based on your specific prompt, and so on and so forth.
It would take a team of developers to create a system that could do all of that. The challenge of letting a generalist artificial intelligence in the homes of random consumers is far more difficult.
What if DeepMind succeeds? What if the world's first artificial intelligence assistant capable of assisting in your day-to-day life is the Google Assistant?
Everyone is going to forget about the virtual assistant if DeepMind and Google can turn it into something that looks like a real life assistant. And, of course, the voice of the people. GATO can do what any other assistant can.
I'm not sure DeepMind can pull it off, but I'm pretty sure the odds increased when the company signed a contract with the GANfather.
Who knows what happened at Apple at the end of the day. Maybe Goodfellow wasn't happy, or maybe Apple wasn't happy.
Most of what everyone is trying to accomplish in the field deals with some of Goodfellow.
It's also worth mentioning that big tech always tries to lure talent from each other. Goodfellow joined OpenAI and then joined Apple twice.
The timing of his joining the DeepMind team is exciting. He is an independent researcher. He will be given everything he needs to do his best work.
Tim Cook may have good reasons for letting his star quarterback leave to join a rival team in the middle of the playoffs. It's hard to see from our point of view, but it's possible.
It is an exciting time for AGI research. There is no telling what Goodfellow and the DeepMind team can do.
Make sure to sign up for the Neural newsletter so you don't miss anything.
All Rights Reserved © 2024
