In the first part of our blog, we got to know the technology behind our digital assistants and took a closer look at how AI works. We also discussed the challenge of making unstructured data accessible to intelligent systems to enable the processing and understanding of information within them in such a way that cognitive processes similar to those of humans become a real possibility. In theory at least, this would result in a truly intelligent system. In part two, we want to dig a little deeper into the limitations of AI for digital assistants and analyse the research status quo.
Let’s first cast an eye over the German research initiative, Humane AI, which revolves around the use of artificial intelligence in digital assistant systems. A group of renowned European scientists have written a detailed account of the challenges facing the creation an artificial, yet human-like system with the aim of developing AI technology that enables machines to act more like people and work in partnership with them to enhance human abilities and empower society.
The authors note that these kind of systems need to be able to learn, reason and plan. To enable collaboration with their human partners, these intelligent systems must not only be able to provide explanations at the end of the learning or reasoning task. They must also be able to give feedback on progress and be able to integrate high-level human input.
Secondly, human-aware interaction and collaboration will require multi-modal perception of dynamic real-world environments and social settings (including perceiving and interpreting related emotions, motivations and social structures) including the ability to build and maintain comprehensive models of environments and of humans interacting within such environments. Intelligent systems must share an understanding of a problem’s larger context to properly cooperate in developing a solution with their human partners.
The research paper also discussed how these aspects should be approached. There has already been some success in developing self-learning speech systems with which people can converse.
However, events from 2016 show that these systems are a long way off being perfect. The Tay chatbot was developed by Microsoft and launched on Twitter, Facebook, Instagram and Snapchat. The aim was to learn from her interactions with other users and become more intelligent doing so. Microsoft was forced to shut down the service after only a few hours as Tay had begun to post inflammatory and offensive messages as the system was unable to differentiate between what is good and bad or acceptable and unacceptable.
These kind of limitations demonstrate that there is still a long way to go before a chatbot will be able to piece pieces of information together in such a way that even more complex processes can be automated. As things stand to today, the idea that there will ever be an autonomous, intelligent chatbot that can work on unique and innovative projects without the support of humans is still very much pie in the sky. This is first and foremost down to the fact that companies’ working environments and data aren’t sufficiently structured as they have grown over time.
The first step on the long road towards fully-automated digital assistants is to prepare corporate environments so they can deliver the type of structured data a machine needs to be able to make predictions. Currently, most relevant information is in employees’ heads. Some of it is stored as data, but not necessarily completely accessible in a structured or digital form in the working environment.
Another issue getting in the way of complete automation is the complexity of the environment a machine has to keep in mind when making decisions. While a self-driving vehicle basically only has to monitor the road ahead, a digital assistant needs to have a constant overview of the entire environment.
As progress is made in this respect, questions concerning AI technology’s reliability, acceptance, transparency and not least legal, ethical and moral responsibility are also being raised. While these issues have been debated for many years in some application areas, in terms of business use, the topic is only just being talked about, but many of them have already been pondered by information ethicists.
Let’s not forget, as AI continues to develop, the workplace of the future will probably be even more strongly characterised by collaboration between man and machine. This makes it worthwhile for companies to look into how chatbots can be used in daily business with a view to automating existing work and business processes.
Project management is a particular area of interest as of 1,800 projects mangers surveyed, 54% said they spent most of their working day dealing with administrative coordination and control tasks such as appointment planning, resource distribution and reporting, with not much time left for important topics such as people development and strategy and innovation. From a business perspective, project managers need to focus more of their efforts on driving business growth and leave routine tasks to an assistant. With the development of AI solutions such as chatbots, this could well be the case in the next decade. In part three, we’ll take an in-depth look at the areas of application and potential chatbots offer in the field of project management.