Today, hundreds of companies are working on machine intelligence capabilities that augment human knowledge-building activities. These efforts are primarily
addressing three major problems. One is information overload, which is exacerbated by ever-increasing volumes of unstructured data including text, audio,
and video (e.g., Big Data). Another is the barrier to efficient communication between people from different disciplines and professions during collaborative
sensemaking. The third major issue is the training and evolution of intelligent machines. HyLighter is a uniquely capable Social Machine (i.e., a general
model for unifying machines and social processes) that addresses these problems in the context of collaborative sensemaking involving large collections of
online documents (i.e., multi-document sensemaking).
Multi-document sensemaking activities require individuals and groups to combine pieces of information from online sources to produce something of value. In my
view, informed by over 20 years in the field of information technology, HyLighter is the most effective platform in the world today for bridging the gap between
smart machines and non-technical users of machine observations engaged in multi-document sensemaking. Arguably, most technologists have not even imagined that a
tool like HyLighter can exist, let alone imagined its potential impact. Nonetheless, initial feedback from a number of providers and users of AI and related
technologies have confirmed our view that HyLighter stands alone in filling this critical gap in technology.
The core concept of the HyLighter vision is the Conversation Model – a simple data model that defines three concepts: the HyLight, the Conversation, and the
Participant. For our purposes, a HyLight is a fragment of information with “point at/point of” attributes, a Conversation is an aggregation of HyLights, and
a Participant is an author and/or consumer of HyLights in a Conversation. To elaborate, in creating a HyLight, the Participant (i.e., human or machine)
defines an addressable fragment of an online source and adds observations as meta-content (i.e., content about content). In other words, the Participant
points at a fragment and says why the fragment is relevant to a topic of interest (i.e. point at/point of). A Conversation can be as simple as a search
result from Google or as complex as a Knowledge Trail (i.e., a preferred sequence of HyLights through a collection of documents that informs the reader by
telling a story). Participants in Conversations can include both humans and machines in the roles of Speakers (i.e., users who can create, modify and arrange
HyLights in Conversations) and/or Listeners (i.e., users who can read Conversations).
The Conversation model creates the foundation for a rich ecosystem of machine/human social interaction with online content. In practice, users create HyLights by
pointing at self-selected fragments in online documents and adding information that elaborates on the point of the fragment. Users create Conversations of value
by selecting HyLights that are relevant to a topic of interest from multiple online sources and arranging the HyLights into a meaningful sequence. In addition,
with machines as Listeners to human feedback represented in Knowledge Trails, we can provide an efficient process for training our AI teammates.
The HyLighter vision requires an architecture that can scale to Google-sized volumes of information. To date, we have focused on implementing core fundamentals of
data composition, flowing data, and functional principles in the browser application. As a result, we have seen vast improvements in the performance and capabilities
of the interface. In particular, the new mashup feature, which was the result of an extraordinary emergent design process, is the first realization of the Conversation
model. The next step will enable both machines and humans to engage in sensemaking activities that augment human intelligence and improve the capacity of machines to
assist humans in multi-document sensemaking activities and creative processes.
In sum, a gap exists in the application of machine intelligence in the market today. From a personal perspective, I see improved multi-document sensemaking through the
use of HyLighter capabilities as the next step in the evolution of augmented human capabilities. My commitment to HyLighter derives from my belief that we are building
the foundation for the next stage in the evolution of information technology by bridging the gap between smart machines and humans. The unique capabilities of HyLighter
(with patent pending status) have the potential to make an enormous impact on organizations, even in this early stage of HyLighter as a commercial product. With a release
imminent, the new HyLighter will empower people from all walks of life to engage in analysis and synthesis activities at scales and speeds unseen in human history. I believe
the long-term impact of Social Machines like HyLighter will change human life as much as the Internet has in the last 30 years.