iF Quarterly: AI & Us
Intentional Futures
A research, design, and strategy consultancy solving hard problems that matter
How iFsters and our clients are incorporating generative AI in their work
It’s an exciting time to have access to the internet – (mostly) unfettered access to the wide, exponentially growing world of generative AI has rapidly changed the landscape of what is possible in industries ranging from healthcare to social impact to sustainability. Significant advances in large language models paired with a radical shift in accessibility has enabled individuals everywhere to unleash their creativity and explore generative AI's potential with little to zero knowledge of the technology driving it.
Our work at iF spans a similar gamut of industry, and in most client engagements AI is bubbling to the forefront of our partners’ minds. Where does generative AI hold the greatest promise? Where does it introduce new risks? How can we accelerate our learning and experimentation
Concurrent to client work and general knowledge sharing internally, the Digital Strategy & Innovation team recently concluded an organization-wide experiment to document how iFsters are using AI and for what purposes. The information below is just a small snippet of the many uses iFsters are finding for generative AI in their everyday work.
In the spirit of the theme of this edition of iF Quarterly, the headshots featured below were generated using a combination of iF headshots and Stable Diffusion, an open source generative AI tool. To address privacy concerns, Stable Diffusion was downloaded to an iFsters’ personal laptop, so no personal information - including imagery - was uploaded to the web.
How we're being responsible, safe, & curious
Just like in our project work at iF, we take a learning disposition in our approach to exploring and testing new technologies – including generative AI (genAI). While there are plenty of reasons to be curious and excited about genAI, there are also reasons to be cautious. Some of the things that we’ve been paying close attention to are: Data accuracy
We’re doing a few things as we begin exploring genAI to maintain the highest standards of our work, delight clients, and shape the world for the better:
Data Quality – When researching or analyzing data, we extract and validate sources, check relevancy and fidelity, and cross-check against multiple sources. If we can’t validate something, we do not include it in our work.
Bias Potential – We engage in training and education on how to identify bias that genAI may include in responses, and strive to take an inclusive, equitable approach
Human Ownership – Because large language models (LLMs) are changing daily and ultimately are still machines, a skilled, trained, compassionate set of human eyes
We’re learning about this exciting new technology more and more, but we acknowledge our responsibility to do so in safe, ethical, accurate ways – for our staff, our clients and partners, and our communities.
How iFsters are using generative AI
Tynan Gable, Lead Strategist
“I use Bard to validate basic research findings
Peter Arbaugh, Technical Product Manager
“I use ChatGPT constantly when I’m working with code. It’s a great tutor and helps me spot simple bugs, understand errors, and explore different approaches.
Olivia Rao, Project Manager
“I've used ChatPDF to help me sift through many long documents which has saved me time by summarizing research and answering my questions to help direct my time and attention to the parts that are most relevant to my work.”
Recommended by LinkedIn
Gentry Croley, Senior Strategist
“I always have ChatGPT open in my browser. I mostly rely on it to unearth and synthesize vast amounts of data and information, but I’ve also found the model to be a great ‘thought’ partner for more strategic efforts.”
Malia Nakamura, Digital Marketer
“I use ChatGPT to help me generate marketing content fairly regularly, and I also use Read.ai to summarize meeting transcriptions and capture action items without being distracted by taking notes.”
Sam Bazant, Lead Strategist
“I mostly use ChatGPT for idea generation
How our clients are using generative AI
Likewise developed a ChatGPT plug-in, and they’re exploring further uses of generative AI to enhance their users’ experiences by coupling GPT with tens of millions of data points to make the process of discovering new content more engaging, personalized, and efficient.
Quill.org collaborated with CommonLit on a free new tool that allows educators to determine whether a text passage was created by humans or AI.
Happenings
Other things we’ve been working on...
Jackie, Christian, Jen, and Olivia are working with the local government’s Safe and Thriving Communities Division to assist in the development of a Victim Compensation Fund program. The goal for this program is to be eventually housed outside of the City. The program is intended to reduce the disparate racial outcomes, shrink and reduce the harms of the criminal legal system, and provide an alternative response to police intervention.
Peter, Eric, Jeff, Andrew, and Michael are imagining how generative AI can be implemented into a media recommendation platform to improve and elaborate on the user experience.
Read, watch & listen
We asked iFsters “What’s something you read, saw, or heard this past quarter that gave you a practical, actionable takeaway?”
Andrew Phyo, Strategic Technologist – OpenAI: Forecasting potential misuses of language models for disinformation campaigns and how to reduce risk
My biggest takeaways from this report summary are 1) how generative language models may be leveraged by propagandists to power influence operations (e.g. disinformation campaigns) on a wider scale than ever seen before, and 2) how we can reduce the threat of influence operations going forward by weighing and implementing various mitigation strategies. I suspect developers, organizations, and governments will have to move quickly and carefully to help mitigate the threat of AI-enabled influence operations in future elections throughout the world - without stifling generative AI's myriad innovative good-faith use cases.
ChatGPT, most popular AI tool at iF – Planet Money: AI Podcast 3.0: Dial M for Mechanization
In this podcast episode, it is demonstrated that AI can be trained to write original scripts by using real research and interviews as input. Additionally, the AI technology is capable of cloning the voice of specific individuals, as exemplified by the replication of the former colleague Robert Smith's voice. By combining the AI-generated script with the cloned voice, a 15-minute episode of Planet Money is created. The performance of the AI is then evaluated to determine its strengths and weaknesses.
David Lickwala, Accounting Specialist – Business Inside: America is Entering a Forever Labor Shortage
This article sought to investigate the labor shortage in the United States, and answer the following questions: What effect will this have on the wage-price spiral and inflation? How can employers adapt to increased wages without raising prices?