Predii Industry Update December 2022
Key Industry Insights December 2022
1. Automotive-specific NLP Language Models are becoming a core element inside enterprises, driving internal and external applications in customer support, after-sales workflows and predictive analytics use cases (more Details below).
2. An economically challenging year 2022 closes with organizations shifting budgets for Data and AI and actively planning for Horizon 3 use-cases for next generation automotive service.
3. Newly emerging key data products include customer-facing ‘concierge’ service models, as well as supply chain optimization: ‘Concierge’ service models leverage AI to power trust-centric applications as consumers have increased access to vehicle service and maintenance analytics, both of their own personal vehicle & from aggregated localized data across other same YMME. On the OEM side, predictive analytics products, e.g. Breakdown Prediction, complete a customer experience-centric product ecosystem. In the aftermarket, emerging data products answer to the needs of an educated, informed customer base and are an opportunity to maintain advantages of trust, price, and convenience. Predictive Analytics is taking the aftermarket from ‘just in time’ inventory planning to ‘ahead of time’ analytics. AI-powered supply chain optimization enables ‘right part, right place, ahead of time™’. planning.
4. AI-powered solutions disrupt traditional business models and competitor landscapes. We see the Roadmap for AI in Service rapidly evolving around Horizon 1, 2, and 3 use cases. With increased access to and AI-enabled actionability of consumer data, aftermarket providers branch out and add previously unrelated customer experience products to their core business model.
Which trends are we observing specifically in Natural Language Processing?
1. Industry-funded research drives specialized, applied language models: Large language models (LMs) have rapidly become a mainstay in Natural Language Processing. These models are known to acquire rich linguistic knowledge from training on large amounts of text. In recent years, LLMs, deep learning models that have been trained on vast amounts of text, have shown remarkable performance on several benchmarks that are meant to measure language understanding. Unlike traditional AI research funded either by the government institutions like DARPA, DOD or originating from the academic circles, most recent breakthroughs in LLMs are originated and funded by the industry: All recent major breakthroughs came from industry (Transformers, BERT from Google, GPT from OpenAI, DALL-E from OpenAI, Galactica from Facebook, etc.)
2. Domain-specific Language Models in enterprises are successful when based on a carefully orchestrated roadmap: A typical roadmap for an enterprise getting into deploying AI Models powered enterprise applications includes: Understand the problem (customer needs) & understand the data (very critical)
Generate synthetic “gold-standard” data or annotate a seed set, pre-processing the data to suit the desired outcome
Select a LM for the task in hand
Perform task-specific fine-tuning, post-processing
Test on real data
Repeat the loop and perform extensive QCTraditional feature engineering/ablation study would be the helpful baselines to understand the quality of predictions.
3. Generative AI is the talk of the town - and it‘s here to stay: a recent example is ChatGPT, a generative AI application based on large language model trained by OpenAI. However, as with any AI application, it is most impactful when it augments human-driven content creation with the percentage of human involvement varying by the application of the content. Read more at in this HBR article.
We‘ve given ChatGPT a couple of test runs - here‘s what we asked. The responses are entirely generated by Generative AI.
(Our question) What is the business value of generative AI?
(ChatGPT) “The business value of generative AI lies in its ability to improve efficiency and accuracy, as well as its ability to generate unique and high-quality content. For businesses and professionals that create a lot of content, such as marketing materials or social media posts, generative AI can be a valuable tool for quickly and easily generating a large amount of unique content. This can save time and effort, allowing businesses to focus on other important tasks. Additionally, generative AI can help to improve the accuracy and quality of the content that is produced, as it can assist with tasks such as summarization, translation, and data augmentation. Overall, the use of generative AI can help businesses to improve their productivity, efficiency, and content quality, which can ultimately lead to increased revenues and a competitive advantage in the marketplace.”
(Our question): What are the main challenges in automotive service?
(ChatGPT) „There are many challenges in the automotive service industry, including:
The need to keep up with rapidly changing technology, including advances in vehicle design, materials, and manufacturing processes.
The need to provide high-quality service in a cost-effective manner, while also ensuring that vehicles are safe and reliable.
The need to attract and retain skilled technicians and other workers, who may be in high demand in other industries.
The need to manage complex supply chains and inventory systems in order to ensure that the right parts and tools are available when needed.
The need to comply with a wide range of regulations, including environmental regulations and safety standards.
The need to adapt to changing consumer preferences, including the increasing demand for electric and autonomous vehicles.
The need to compete with other automotive service providers, including independent shops and dealerships, who may offer similar services at lower prices.“
Predii Thought Leadership on 2022 Industry Events
How AI is disrupting the Service Ecosystem [Automotive Service Symposium, New York City] Understanding the Automotive Data Ecosystem today and in the future [2022 Jefferies Automotive Aftermarket Investor Conference, NYC] Trending Use Cases for applied AI in Automotive [IOT Solutions World Congress 2022, Barcelona] Challenges with Applied AI in Automotive [IOT Solutions World Congress 2022, Barcelona] State-of-the Art NLP Trends [ISO/IEC AI Workshop Series Emerging Tech Trends, virtual] If you are interested to learn more or if you would like a copy of our presentations, please feel free to reach out.
Comments