Facebook arbeitet an einer KI, die automatisch Code von einer Programmiersprache in eine andere übersetzen kann. Das Feld ist nicht gänzlich neu, es gab schon Versuche, KI zu verwenden um beispielsweise statische Analyse und damit Code-Qualität zu verbessern. Wenn eine Maschine zwischen Code übersetzen kann, werden eine ganze Reihe von Banken jubeln, endlich nicht mehr auf unersetzbare, uralte Cobol Programmierer angewiesen zu sein, die auch noch unfassbare Stundensätze aufrufen.
Es ist eine Gelegenheit, schwierigere Probleme anzugehen.
TensorFlow, the open-source machine-learning library, introduced a library to make recommendations easier. Recommendations are a crucial component for e-Commerce but also other web-services. Good recommendations help build a better user experience and drive customer engagement. The more time consumers spend on a site, the quicker customers find what they are looking for, the better their satisfaction. Recommendations help achieve this and TensorFlow now makes it easier to improve such functionality.
Introducing TensorFlow Recommenders, a library for building flexible and powerful recommender models.
It’s not like email has been a perfect solution ever, to start with. In fact, email has been broken for most of its existence. Imagine all the rules and filters you need to stay on top of your inbox. When Internet became popular, soon spam became popular.
Email lists were usable only before eternal September began. Just the other day somebody at my employer responded to an email list that has thousands of subscribers. And so did everybody else.
Not to mention those emails that come with a good intention and make it past all filters into your inbox. Those typically span many pages and make you feel guilty for not reading because you’re busy.
Admit it, email is broken.
With the advent of new technology, there are new solutions. Magic Email, as announced by Producthunt in it’s weekly newsletter, is something that I’m totally not sure whether it’s an improvement or total troll. Built on GPT-3, that created some bus on the Internet recently, Magic Email allows you to do two things:
a) It will summarize long emails for you. That actually seems to be a good idea for those emails you just couldn’t get around to reading full detail. At least it will tell you whether it’s worth it to invest more time and go into the details buried in long prose.
b) The much more interesting feature it is, that magic email can write text for you. You just give it a bunch of keywords a simple statement and it will extend to a long email. This is exactly that part that makes me wonder whether the product is meant as a troll. Imagine all those guys responding to an email list. Instead of replying “please unsubscribe me”, GPT-3 will write an exhaustive email basically saying the same. The same level of detail will pull much more of your time. Unless you have Magic Email installed yourself of course.
Nevertheless, the product is amazing.
Magic Email is your AI-powered email assistant that summarizes your emails and generates professional emails from brief one-line descriptions. Get through all of your emails 5x faster so you can free up more time for your important work.
AI didn’t get smarter: GPT-3 has gained a lot of online attention in the past couple of weeks. GPT-3 is short for Generative Pretrained Transformer 3, that – in short – produces text. It is a commercial offering from OpenAI, a company to drive forward peacefully usage of artificial intelligence. Among others, Elon Musk is one of the founders. However, the company and it’s founder parted ways, citing “a potential future conflict (of interest)“.
Liam Porr, an AI developer, published an article in which he describes how he ran a successful blog with that GPT-3 technology. Key takeaway:
I would write the title and introduction, add a photo, and let GPT-3 do the rest. The blog has had over 26 thousand visitors, and we now have about 60 loyal subscribers… And only ONE PERSON has noticed it was written by GPT-3.
from the blog.
The first post even made it to #1 on HackerNews. The article reads pretty natural, but fails to make a solid argument. Content-wise, I’d take this as indication that not AI got smarter, but humans got dumber. The particular article from above could’ve been written by a random hipster working in content marketing.
Japanese researches looked into a recent Tesla Model 3. Their analysis has some interesting findings over traditional car manufacturing methods. In particular when it comes to electronics, “ECUs” how they’re called in the automotive world, “Electronical Control Units” A regular Toyota or European car relies on dozens or more of these to make the car work.
However, research found that Tesla really only relies on one central component to take care of both autonomous driving and the entertainment part.
Self-driving AI sends shivers through traditional supply chains
From the article
This actually is big news, because it indicates Tesla has chosen to develop core technology in house, becoming (more) independent of supply chains. As of this writing, Tesla produces a fraction of what VW and Toyota output. To achieve the scale, automotive Industries traditionally groomed a rich ecosystem of suppliers, to form the entire value chain.
Automakers worry that […] will render obsolete the parts supply chains they have cultivated over decades, […]
From the article
However, it appears Tesla has an substantial competitive through this supplier independence. All of the above worry aside, Automakers will have to invest their capacity and headcount to catch up with this assumed advantage of 6 years.
Intel acquires Habana Labs: Intel put a heavy bet on Artificial Intelligence today. In a press-release, the company announced the acquisition of Habana Labs, a company founded and based in Caesarea, Israel. The deal was worth “approximately $2 billion”.
To push its artificial intelligence strategy, Santa Clara-based Intel has acquired Israel’s Habana Labs for “approximately $2 billion dollars.”
Technical University Munich Institute for Ethics in Artificial Intelligence launches a speaker series to bring experts from all over the world to Munich and talk about Ethics and Governance for Artificial Intelligence. The Series kicks off with an Inaugural Session with Lionel P. Robert on December 13 – 10:00 am – 11:30 am.
With its new Speaker Series, the TUM Institute for Ethics in Artificial Intelligence is bringing experts from all over the world to Munich to talk about Ethics and Governance of […]