Researchers have developed a new artificial intelligence (AI) framework that excels in analyzing and categorizing conversations between individuals, with the aim of improving team training techniques. This framework will enable training technology to better understand coordination and teamwork among people.
"People are very interested in developing AI training technologies that can understand the dynamics of teamwork and modify their training to promote better collaboration among team members," said Wookhee Min, a research scientist at North Carolina State University and co-author of the research paper. "However, previous AI architectures have struggled with accurately assessing the content shared by team members during communication."
"We have developed a new framework that significantly enhances the AI's ability to analyze communication between team members," said Jay Pande, a doctoral student at North Carolina State University and the first author of the paper. "This is an important step in developing adaptive training technologies aimed at facilitating effective team communication and collaboration."
The new AI framework is built upon a powerful deep learning model trained on a large text-based language dataset. This model, called Text-to-Text Transfer Transformer (T5), was then customized using data collected during training exercises conducted by the U.S. Army.
"We modified the T5 model to more accurately analyze team communication using contextual features, such as the role of the speaker," said Min. "This background information can be crucial. For example, what the team leader says may need to be interpreted differently from what another team member says."
To test the performance of the new framework, researchers compared it with two previous AI technologies. Specifically, they tested the abilities of these three AI technologies to understand the conversations of six soldier groups during training exercises.
The AI framework has two tasks: classifying ongoing conversations and tracking information flow within the team. Classifying conversations involves determining the purpose of the spoken content, such as whether someone is requesting information, providing information, or giving commands. Tracking information flow refers to how information is shared within the team, such as whether it is passed up the chain of command or down.
"We found that the performance of the new framework was significantly better than previous AI technologies," said Pande.
"One particularly exciting aspect is that we trained our framework using data from one training task, but tested the model's performance using data from another training task," said Min. "Despite testing the model in a new environment, the performance improvement compared to previous AI models was significant."
The researchers also noted that they used a relatively smaller version of the T5 model to achieve these results. This is important because it means they can obtain analysis results within seconds without the need for supercomputers.
"The next steps include exploring the applicability of the new framework in other training scenarios," said Pande.
"We tested the framework using transcribed text data from audio files of training sessions," said Min. "The next step will involve integrating this framework with AI models that can transcribe audio data into text, in order to assess the real-time capability of this technology in analyzing team communication data. This may involve improving the framework's ability to handle noise and errors, as AI transcribes audio data into text."
This paper, titled "Powerful Team Communication Analysis Based on Transformer Dialogue Modeling," will be presented at the 24th International Conference on Artificial Intelligence in Education (AIED 2023) in Tokyo, Japan, from July 3rd to 7th, 2023.
Source: https://techxplore.com/news/2023-06-ai-boosts-teamwork.html
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.