Background
While there is a lot happening in the world around us that’s dominating the airwaves, the march of progress on bleeding edge computing and AI has continued. Although there have been disappointments [1], we’ve been paying close attention to a few high profile and important developments [2]. Google, DeepMind, and OpenAI have shared incredible results related to their Large Language Models (LLMs), in addition to OpenAI’s DALL-E 2 that generates images based on text inputs using deep learning via transformers. Per OpenAI, it can understand the relationship between two different objects and combine them, such as “teddy bears mixing sparkling chemicals as mad scientists” in photo realistic quality – see first photo in post.
See several examples of DALL-E 2 from this Twitter thread.
“Teddy bears mixing sparkling chemicals as mad scientists” – Generated by DALL-E 2 from text prompts (Source)
The results of this technology are astounding given the newness of transformers (circa 2017 [3]) and recent take off of deep learning generally (circa 2015 [4]). And these technologies will have broad implications to many aspects of our lives as the technology becomes more affordable and packaged into products by Software-as-a-Service companies and initiatives (e.g. Cohere, Hugging Face, EleutherAI etc.). However, most organizers will be hamstrung by inconsistent data, lack of governance, and baseline customer data management approaches. Credo AI is one organization solving for one of these areas [5]. It’s not enough for your organization to “keep the lights on” today.
Implication for Marketing & Advertising
We see major future implications for Marketers and Advertisers:
- Content and Copy Development – Coded and personalized for specific audiences or audience attributes. It is not a far off world to ask a Marketing AI to, “Give me 5 versions of an email that is most persuasive to someone who has recently browsed my website (and looked at sandals, lives in the Chicago area, and hasn’t bought from us since before the pandemic).”
- “You can now give an AI system like GPT-3 [a LLM] the text: ‘ I’m going to go to this mountain with the faces on it. What is the capital of the state to the east of the state that that’s in?’ And it gives the correct answer (Saint Paul, Minnesota) — something most AI researchers would have said was impossible without fundamental breakthroughs just seven years ago.” per Dr. Max Tegmark of MIT and Future of Life Institute in a recent interview. [6]
- Creative copy development for personalized advertisements, such as “one liners” for optimal direct response advertisements.
- Chatbots and customer journey management – The ability to combine chatbot technology with the most optimal customer journey – for both your organization and the individual that’s interacting with your brand.
The most important input to fuel these technologies is clean, labeled, and well-governed customer data. Is your team and organization prepared? Reach out to the two authors to talk more about customer data governance, data strategy, implementing large language models, and customer journey analytics.
Sources:
[Image Credit] https://ai.googleblog.com/2022/04/pathways-language-model-palm-scaling-to.html
[1] https://hbr.org/2022/03/why-ai-failed-to-live-up-to-its-potential-during-the-pandemic
[2] https://www.nytimes.com/2022/04/15/magazine/ai-language.html
[3] https://arxiv.org/abs/1706.03762
[4] https://en.wikipedia.org/wiki/Timeline_of_artificial_intelligence
[5] https://www.credo.ai/
[6] https://80000hours.org/podcast/episodes/max-tegmark-ai-and-algorithmic-news-selection/
Transparent Partners
