Alexandria Smith – Sales & Marketing Intern
Sandline remains at the cutting edge of technology and as GenAI becomes embedded in every corner of the digital landscape, it’s important to consider all aspects. Not just what AI can do, but what it costs to get there. That means looking beyond efficiency alone and considering the benefits as well as the ethical and security implications of how we use these tools.
Sandline’s mindful, measured approach to AI isn’t just good practice, it’s the future of responsible legal tech. Read on to see how we’re making AI work smarter for our clients, and for the world it touches.
Security
Many AI tools, particularly large commercial models, both open-source and closed- source models, may outsource data processing and training to offshore locations with weaker security standards. This practice raises serious concerns about client confidentiality and data privacy.
It’s important to distinguish between open source, closed source, and the secure loop process.
Open-source Large Language Models (LLM’s) make their code and models freely available to all. They can be customized and deployed as desired. In addition, because they do not entail an overhead cost, they offer significant cost savings.
Closed-source LLM’s, like ChatGPT are entirely private; their model, architecture and training data are not publicly available. They must be accessed through their own server endpoints, at a fixed cost. In addition, these models may use client prompts to train their models, which raises potential security and privacy concerns.
The secure loop process allows for fully self-contained systems. All data remains within a secure environment, and the model does not use your inputs to alter the model or improve other systems. Within a secure loop process, you narrow the scope, and the AI only responds to what it is directly given. For example, if you input case documents the AI wouldn’t source information about the company or individuals, it would only look to find what you had asked for. Every interaction is monitored, controlled, and never shared with third parties. This makes them ideal for the legal services field where privacy is non-negotiable. Thats why Sandline takes a dedicated approach that prioritizes security using AI only when it is necessary and uses smart AI workflows to the discovery process.
Ethics
Having a mindful approach to AI is not only helpful to optimize workflows but necessary given the environmental consequences of AI usage. AI is not a magical black box, even if it is called “cloud computing” that doesn’t mean the data centers aren’t present in our world. Training AI models takes a lot of electricity and thus indirectly a lot of water to cool down the systems. Every time a model is used, even just asking Chat GPT to summarize an email, this operation consumes energy. Researchers estimate that ChatGPT inputs consume 5 times more electricity than a simple internet search1. The electrical demands of generative AI interfaces become increasingly complex as these models become larger and the application of them is applied ubiquitously.
The direct impact of these electricity demands at the physical data centers leads to large consumption of water in these facilities. Chilled water is used to cool data centers by absorbing the heat from computing equipment. Bashir, a Computing and Climate Impact Fellow at MIT Climate and Sustainability Consortium (MCSC) explains that “each kilowatt hour of energy a data center consumes, it would need two liters of water for cooling.”2
Other studies have found that ChatGPT uses 39.16 million gallons a day internally and 14.28 billion a year3. The energy and water usage of AI is projected to increase rapidly by 20284.
Why do we tell you this? Because we understand that AI comes at a cost, not just financially, but environmentally and ethically. That’s why Sandline is committed to having a highly trained team that understands the full system behind Generative AI. Our experts know how to optimize these models, decreasing run times, and reducing unnecessary energy usage.
Precision in prompt engineering isn’t just about efficiency it’s a matter of reducing waste. The more refined your prompts, the fewer iterations are needed, which minimizes both security vulnerabilities and the social or ethical impacts of data processing.
When used intentionally, AI can be both powerful and sustainable. At Sandline, we make sure it is minimizing environmental and security impacts while delivering real value to our clients.
The Sandline Approach
At Sandline, our approach to AI begins where it matters most: with security. Before we consider implementation, we evaluate whether AI is necessary and beneficial for the task at hand. We move forward when it can meaningfully improve workflows while maintaining the highest standard of privacy and defensibility. We do not limit ourselves; we look at all options broadly and thoroughly.
Sandline is known for its innovation. Our process is built on constant touchpoints and real time feedback loops. We continuously test existing tools and develop our own to solve modern data challenges. As AI technologies evolve, so do we. We’re not just adopting technology; we’re building the future of legal tech in the AI landscape.
We don’t take a one-size-fits-all approach. Instead, our consultants and advisors work closely with clients to assess a wide range of options: balancing cost, compliance and performance. We’re not limited to tools like ChatGPT. We explore, select and build the most effective solutions for every project, always with a thorough understanding of each model’s capabilities and risks.
Conclusion
When you understand the models, you’re working with your results aren’t just accurate, they’re defensible and valid with the ethical standards legal services demand, that is the Sandline way.
