Articles > AI for Business - Your 5 AI reads of the week #12

20 days ago


alt text

Hello readers and welcome back to our weekly AI for Business reads. Every Thursday, we share our weekly selection of interesting AI for Business reads. If you missed some of our earlier must-read lists, head over to our Articles section to catch up!

We are happy to see more and more traction towards our AI for Business reads, hope you will also enjoy this week’s list!

In this week's edition, we will take a closer look at:

  • Autonomous navigation system
  • The challenges of rescue drones
  • How emotion is a new metric for advertisements
  • Using AI to understand the news
  • How AI wants to know how you plan a trip

Enjoy!

FIGURE OF THE WEEK

According to the McKinsey Global Institute, machine learning and artificial intelligence are set to transform the banking industry with a value of more than $250 billion. ML and AI models will be able to easily crunch the vast amounts of data which will improve the decision making, tailor services and risk management.

HUMAN REASONING WITHIN AI MODELS

MIT researchers wrote a paper about their new autonomous navigation system that should enable a car to drive autonomously thanks to imitating human steering behaviour. The human feature that machines are missing (for now) is human reasoning: being able to drive on a road that you haven't seen before thanks to comparing it to other roads.

The model of these researchers uses a branch of machine learning called convolutional neural network (CNN) which is commonly used for image recognition. This model is an end-to-end navigation system that has matched every steering command a human makes at any given instant. This should enable the car to drive on a road he has not driven on before.

When finding missing people, speed is of the essence! Coordinators are continuously looking for new tools to help in this endeavour: volunteers, scent-trained dogs, vehicles etc. With drones and AI becoming proven new tech, research is being done to help rescuers.

But there are still a couple of challenges that need to be tackled: drones gather a lot of data, lack human reasoning and still need a lot of human support. Researchers from Virginia Tech are trying to solve these challenges by implementing machine learning into the drone. It will be able to recognize the differences in heating signature between an animal and a human. This ML model will crunch the data and give results to the ground forces. They are also adding human knowledge, for example how a child thinks when running away compared to someone with dementia.

Very promising evolutions in the search for the perfect rescue drone!

The New York Times has been using machine learning last year to predict the emotions felt by readers of various stories. Recently they created a list of 30 emotions that were often experienced, of which 18 can be commercialised. Meaning that advertisements can now be placed based on the emotions felt by the reader on an article.

The idea behind this is that emotion is seen as additional, easy to understand metric and thus a metric to decide how to approach them. Machine Learning professionals and advertising professionals, tell us your opinion about this !

AI IN PRACTICE

AI can tremendously help in automatically understanding, labelling and captioning images.

Do you know Robert Mueller, who conducted the Special Counsel investigation about Russian interference in the 2016 United States elections and suspicious links between Trump associates and Russian officials? Give to the Google Vision API (i.e solution) some pictures of Robert Mueller and they will directly recognize him, even without using facial recognition. How does it work? When giving a video to Google's Web Entities feature, it performs a reverse Google Image search of the open Web to analyse the content and gives back the most common topics to describe the video. This feature can recognize someone (example Robert Mueller) in a video without having a facial recognition system built in, this by cross-referencing the image of the person with articles on the web of that person.

Deep learning, this growing area of Machine Learning, has revolutionized the capabilities of machines and companies. They can now easily sift through billions of images a day and label everything they see on this image, whether it be violence, nudity or identifying a text.

Competition in the travel industry is very fierce thus travel companies are continuously looking for new mechanisms to reach their target market. The better the companies know you, the more personalised their messaging to you will be. Hence companies such as Expedia are - very widely - using AI in their solutions to provide even more value to their B2C and B2B customers.

One of these tools is eye-tracking software, this enables them to see where on the screen the customer is looking. Another one is chatbots, this virtual assistant provides answers to frequently asked questions and when the bot does not know the answer it sends it as a message to customer service and they add the answer to the system.

And do not forget to subscribe to our newsletter for your monthly AI dose.

See you next week!