Edition 7: OpenAI Takes on Google Search
Welcome to the AviaryAI Newsletter!
Thanks for joining us as we explore the intersection of GenAI and finance with practical learnings and the latest relevant insights. Let’s get started.
This week you’ll learn:
- Why OpenAI's new search product may disrupt digital strategies
- Microsoft's stance on responsible AI usage in law enforcement
- About an AI breakthrough that may slash drug development costs
- What CU execs need to know about AI compute resources
- How Visa is using GenAI to outsmart fraudsters
Subscribe to the AviaryAI Newsletter here.
GENERATIVE AI THIS WEEK
The coolest things we're watching and why you should care
OpenAI to Unveil AI Search Product, Intensifying Google Rivalry
OpenAI, the Microsoft-backed AI company behind ChatGPT, plans to announce an AI-powered search product on Monday, according to sources. The move positions OpenAI to directly compete with Google in the search market and coincides with the start of Google's annual I/O conference, where AI-related products are expected to take center stage.
So what? AI-enhanced search could revolutionize how members find and engage with financial information online. Credit unions may need to adapt their digital strategies to ensure they remain visible and relevant in an AI-driven search landscape. Read the full story here
Microsoft Bans Police from Using OpenAI for Facial Recognition
Microsoft has updated its Azure OpenAI Service usage rules, prohibiting law enforcement from using models like GPT-4 and DALL-E for facial recognition purposes. The ban applies globally to officers using body-worn or dashboard cameras, extending OpenAI's existing restrictions.
So what? As AI becomes more powerful, it's crucial for anyone leveraging AI to stay informed about the ethical implications and potential misuses of the technology. While this ban specifically targets law enforcement, it highlights the importance of responsible AI usage across all industries to protect member privacy and maintain trust. Read the full story here
Google DeepMind's AlphaFold 3 Revolutionizes Biological Research
Google DeepMind has unveiled AlphaFold 3, a groundbreaking AI model that can predict the structures and interactions of nearly all biological molecules, including proteins, DNA, RNA, and potential drug molecules. This landmark innovation, published in Nature, is set to dramatically accelerate biological research by enabling computational testing of molecular interactions, which could otherwise take years of expensive experimental work.
So what? AlphaFold 3 could significantly speed up drug discovery and development. This technology could soon have major implications for your members’ healthcare costs and insurance. Read the full story here
CREDIT UNION AI PROMPT OF THE WEEK
Boost your efficiency with these easy to follow templates
Simplify Complex Documents For Easy Communication
Quickly distill the key points from a complex document into an easy-to-understand summary for your employees.
This could be useful for employee handbook updates, policies, procedures, benefits explanations, or other important documents filled with jargon or legalese.
How to use this prompt:
- Copy and paste the template below into a generative AI model (ChatGPT, Gemini, etc.)
- Replace the [Insert x] with the recommended content
- Do not edit the content within the "{}" brackets.
- Hit “Generate” and review the output
- Ask the AI to make any changes you deem necessary
- Enjoy the time you’ve saved!
--------------------------------------------------------------------------------------You are an AI assistant that helps simplify complex HR documents into easy-to-understand summaries for employees. Your goal is to distill the key points of the document into a concise, clear summary that captures the most important information employees need to know.
Document:
"""
[Paste the full text of the HR document here]
"""
Summarize the document into a clear, concise employee communication following this format:
Document Title: {Provide a simple, descriptive title for the employee communication}
Key Points:
• {Bullet point 1 summarizing a key takeaway}
• {Bullet point 2 summarizing another key point}
• {Bullet point 3 capturing any other critical information}
• {Add additional bullet points as needed}
Required Employee Actions:
• {Bullet point 1 stating any actions the employee needs to take, if applicable}
• {Bullet point 2 with another required action item for employees}
• {Add more bullet points if multiple employee actions are needed}
Additional Resources:
• {Links to relevant policies, FAQs, or contact information for questions}
Please simplify the language to an 8th grade reading level, define any HR/legal jargon or acronyms, and focus the summary on the information most relevant to employees while omitting unnecessary technical details. The tone should be professional but personable.
--------------------------------------------------------------------------------------
Looking for real world use-cases of AI for your credit union? Use our free: AI Prompt Library For Credit Unions
THE AI EXECUTIVE'S HANDBOOK
One simplified GenAI concept per week to build your AI Acumen
AI Compute Simplified - What Execs Need to Know
While credit union leaders may not be building AI models in-house, understanding the relationship between compute and AI is critical for any executive considering an AI strategy. Here’s what you need to know:
AI models are essentially giant math equations that require immense computational resources to solve. GPUs, memory, data storage, cloud resources, are all resources these models need a lot of.
As models grow more sophisticated, the demand for computing power skyrockets. This leads to major challenges:
Reliability - Inconsistent access to compute can grind AI initiatives to a halt. Steady, dependable resources are a must.
Cost - Compute isn't cheap. Expenses can easily spiral out of control without careful planning and management.
Efficiency - Are you getting the most bang for your buck? Optimizing compute is crucial for cost-effective AI.
Your AI Compute Checklist ✅
For credit unions implementing AI solutions, considering asking your AI provider these questions:
- How do our AI providers ensure steady access to compute?
- What steps do they take to optimize efficiency and manage costs?
- Can our AI solutions rapidly scale to meet demand spikes?
- How do compute limitations impact end-user experience?
- What redundancies exist to protect against outages?
- How would compute shortages affect our member service?
AviaryAI’s Approach:
At AviaryAI, we've tackled these challenges head-on. Our proprietary compute infrastructure is optimized for the unique needs of credit unions.
By distributing workloads intelligently and leveraging cutting-edge efficiency techniques, we provide ample compute power that scales seamlessly as your AI usage grows.
And with our private AI models, you maintain total control over member data - it's never exposed to public servers.
The result? Reliable AI solutions that can be deployed quickly to maximize member impact without exorbitant costs.
GENERATIVE AI WORD OF THE DAY
Open Source LLM
Large Language Models (LLMs), are the AI text generation models taking the world by storm. The most popular "Closed" models are the famous ChatGPTs, Claude, and Gemini.
On the other hand you have "Open source" models such as Llama-3 and Mixtral models. The special thing about an LLM being open source is that anyone can freely access, study, and modify the code that makes it work. This allows people to collaborate, improve the programs, and use them for various purposes without restrictions, making them more accessible and adaptable compared to proprietary alternatives.
GENERATIVE AI IN FINANCE
The latest news at the intersection of GenAI and Finance
Visa’s New GenAI Tool Outsmarts Fraudsters
Visa has launched a new tool called Account Attack Intelligence (VAAI) Score to combat enumeration attacks, a sneaky type of financial fraud where bad actors make small test purchases to avoid suspicion before stealing big bucks from unsuspecting victims' accounts. This devious technique costs a whopping $1.1 billion annually!
Visa's generative AI-powered tool learns a cardholder's spending habits and assigns a risk score to each payment in just 4 milliseconds, flagging transactions that look fishy. Trained on over 15 billion transactions, the VAAI Score has already reduced false positive fraud detection rates by an impressive 85% compared to other risk models.
Visa's VAAI Score highlights the growing importance of leveraging advanced technology to combat evolving financial threats. As fraudsters become more sophisticated, financial institutions have responsibility to continually evaluate and improve their fraud prevention strategies to protect their members' financial well-being. While the VAAI Score is just one example of such a tool, implementing such advanced tools can help credit unions maintain member trust, reduce financial losses, and streamline fraud detection processes.
About AviaryAI
AviaryAI is the next evolution of financial interaction. Enhance your team with proactive outbound voice agents to welcome new members, encourage credit card activations, and drive non-interest revenue.
