While artificial intelligence may seem like a new frontier, the reality is many individuals and agencies have been using AI solutions for years without even realizing it. With newer applications such as voice assistants and autonomous vehicles, or less apparent and seemingly mundane use cases such as email spam filters, fraud detection and language translation, AI is likely already assisting your agency.
Artificial intelligence improves public safety’s ability to process data, recognize patterns and improve productivity, and that is why AI has long been used by the team at Axon to enhance and extend the capabilities of public safety professionals.
We recognize the immense promise of ethical AI innovation. Our mission is to harness cutting-edge AI technology to revolutionize community safety, all the while prioritizing the rigorous mitigation of biases and other potential risks. This includes building tools that preserve the crucial role of human decision-making, with specific controls that keeps officers in-the-loop, while significantly improving intervention efficiency and efficacy.
Axon rigorously tests our AI-enabled products to reduce inherent bias in the models before products are released. We are continually monitoring and improving any unforeseen instances of bias that may arise in the models and continue to seek external feedback from Axon’s Ethics and Equity Advisory Council, an independent body that advises Axon on a host of issues through a racial equity and ethics lens, centering the needs and perspectives of communities in the creation of Axon’s products.
So, how does Axon currently use AI, and what are some of the safeguards we’ve designed to ensure AI is used responsibly?
How Axon leverages the power of AI today:
/ Draft One:
Currently, report writing is a large part of each officer’s day. In fact, officers can spend up to 3 hours each day writing reports. Draft One is a force multiplier for officers, leveraging generative AI and body-worn camera audio to produce high-quality draft report narratives in seconds.
Draft One cuts report writing time in half. Report narratives are drafted using the audio transcript from Axon body-worn cameras and are available within 5 minutes of the end of the recording. The underlying model used for Draft One transcription is Open AI’s GPT-4 Turbo, which has been calibrated to prevent speculation or embellishments. Learn more about how Draft One was developed to ensure completeness, neutrality and objectivity in this double-blind study.
Safeguard: Narratives cannot be submitted without officer review and approval. Officers must review the draft, manually fill in missing information and then sign off on the narrative’s accuracy before submission.
/ Axon Auto-Transcribe:
Video and audio evidence are central to every investigation, but the sheer quantity of digital evidence available means officers and prosecutors spend large portions of their day in evidence review. Axon Auto-Transcribe accelerates evidence review and transcription, meaning significant time and cost savings for your agency.
Axon Auto-Transcribe is centered on an AI model called Automated Speech Recognition (ASR). This model intakes audio and outputs text. Auto-Transcribe also powers Dictation, which can enhance efficiency by enabling officers to turn speech into text via their smartphone or computer (similar to how you might dictate a text message through your phone).
Safeguard: Transcripts are labeled “unverified” until an officer reviews, edits and approves the final transcript using our verification tool.
/ Axon Automatic License Plate Recognition:
Automatic License Plate Recognition (ALPR) systems provide high-quality, real-time alerts on plates of interest. ALPR is incorporated in Axon Fleet 3 and Axon Evidence. We use AI models to process raw video frames, detect a license plate, identify characters on the plate and output a string of plate characters to be checked against your agency’s hot-list.
This dynamic technology helps your agency better focus resources and efficiently find offenders and exonerate the innocent. Learn more about how Axon Fleet 3 with ALPR is improving public safety at the Grand Prairie Police Department.
Safeguard: An officer is the final decision maker, verifying the ALPR output before taking any action.
/ Axon Redaction Assistant:
Redaction is an extremely manual process that can take officers hours. Redaction Assistant is a feature of the Axon Evidence Redaction Studio, designed to detect and redact the most common objects found in body-worn videos and save your officers time.
Redaction Assistant utilizes three AI models to automatically detect and redact screens, faces and license plates in video footage. Audio Redaction via Transcript further increases efficiency by redacting the corresponding audio when users click on words in an integrated transcript. All this time saved means your agency can have less officers at the station and more in the field.
Safeguard: Redaction users must review the output before deciding what to blur and verify the redactions before sharing evidence.
/ Priority Ranked Video Audit:
The average agency generates anywhere between 500 to 2,000 body-worn camera (BWC) videos per day. Many agencies require BWC footage be audited by supervisors to review officer behavior, identify coachable moments and inform policy changes.
Priority Ranked Video Audit (PRVA) is part of Axon Performance and leverages AI to proactively flag videos for fast, effective review. PRVA streamlines the BWC review process by surfacing the most critical videos for review, drastically reducing supervisor workload.
Safeguard: PRVA does not utilize AI to assign officers “professionalism scores.” AI simply assists your supervisors by identifying videos that are more likely to need reviewing, but the supervisor must review and make their own assessment about officers’ actions.
/ Axon Fusus:
Our Real-Time Crime Center solution is a force multiplier for agencies that enhances situational awareness and streamlines incident response. The cloud-based platform unifies different data sources into a single pane of glass and leverages AI in a variety of ways.
Specifically, our CORE devices add powerful AI technology directly into existing camera networks, making it possible to search video evidence for vehicles and other objects. Axon Fusus AI also enables users to set alert profiles, providing notifications based on certain criteria. For instance, your agency can receive an alert when a vehicle drives the wrong way or an individual walks into a designated area of interest.
Safeguard: Users must verify the accuracy of automated detections. If a particular object is detected within Axon Fusus, users must confirm before further action is taken. Axon Fusus makes this simple by unifying so many different data sources into a single interface.
What’s next for AI in police work?
AI has already woven itself into the fabric of everyday life. But as we look to the future of public safety, thoughtful and responsible uses of AI can have enormous impact. For Axon, AI isn’t merely a product feature, it’s a catalyst for transformation and we’ll continue to advance AI-enabled applications that increase efficiencies for agencies, enabling officers to spend less time on mundane tasks and more time on core work in their communities while upholding the principles of ethics and fairness.
The use cases for AI technology are rapidly expanding, and Axon is proud to be at the forefront of public safety AI development. Learn more about the role of responsible AI within the Axon Ecosystem.
:format(webp))