In an attempt to enable decision-makers to make more informed decisions, IBM has produced an AI that is capable of taking on (and beating) humans in a debate. Project Debator sifts through vast amounts of information to find relevant points, which it constructs into speech, complete with persuasive rhetoric. It can then listen to and process human speeches, and make rebuttals.
AIs taking on humans have come on leaps and bounds in recent years, but this is a different beast entirely, which makes it all the more impressive. However, there's a chance that training an AI to be persuasive and at times intentionally deceptive (both very human traits), is a field of study that some people will not feel comfortable with.
Google is training machines to predict when hospital patients will die. Doom and gloom aside, the aim is to forecast medical outcomes, including details about what the prediction is based on, so that healthcare providers can improve efficiency. Among other things, the machines will also estimate length of stay and odds of readmission.
It's Google's aptitude for unsupervised machine learning that sits at the heart of this. Having a machine that is able to amalgamate myriad sources of information, including scribbled doctors' notes across multiple files, and make informed decisions based on all the facts could be huge in healthcare – an industry where the margin for error really can be the difference between life and death.
Facebook is currently exploring some interesting ground by offering group admins the option to charge members a monthly fee, in return for which they'll become part of a sub-group and receive exclusive content and connections.
The change was requested by group admins themselves, who have indicated they'd like to be able to invest more in their community, and with a billion people using the groups feature, that's potentially a lot of money. Overall value will be reliant on the 'creator-leader' of the group bringing something special, so it'll be interesting to see how this plays out. Hopefully it won’t downgrade the value of other community members.
Podcasts were huge a few years back, then they went away a bit, and now they're back with a vengeance. As part of the recent boom Google has launched Google Podcasts, the company's first dedicated podcast app since it chose to discontinue Google Listen in 2012.
While not a revolutionary announcement, it is evidence of the resurgence of podcasts and will give a huge amount of people easy access to them. Google has said it wants to double the amount of podcast listeners over the next few years. Well, this is a hell of a good way to start.
US start-up LOOMIA is about to make clothing seriously smart. Not only does the technology contained within its clothes have the ability to emit light and warm you up in the winter, but users (wearers?) will also have the opportunity to sell the data collected via a secure blockchain.
While information such as how many times a garment is worn and the frequency it is washed might not seem interesting to you or me, this kind of insight could be valuable to fashion companies looking to predict future demand. It's great that users are being put in control of their data and hats off to LOOMIA for its cautious approach and use of blockchain to keep data private and encrypted.