Predictive Powerhouse: Elevating AI Accuracy and Relevance with Third-Party Data

Editor's Notes

1. Welcome & Introduce session and Costanza
2. Tendu 1 minute​ Topic: Why​​ ​ ​With such volumes of newly created and at a speed that we have not experienced before, over 90% of business leaders believe that AI is critical to success over the next five years.​ And 75% of enterprises are prioritizing automation, AI and ML and building data science teams.​ ​ So that they can accelerate decision making and business outcomes with more automation, higher scale, predict and create a competitive advantage.​ ​ ​ Forbes 75%​ Deloitte 94%​ ​ ​ FULL STATEMENTS​ 180 ZB Data will be created by 2025​ 94% Business leaders believe AI is critical to success over the next five years​ 75% enterprises are prioritizing AI and ML and hiring data scientist to achieve real results​ 3.5Q Bytes of data created every day​ 200+ ZB data that will be received by public and private cloud storage by 2025
3. Tendu Less than 1 minute​ Topic: Invested in AI​ ​ How many of you are already using AI? See hands and engage audience​ Many of the industries, such as Finserv, Insurance, Retail have been using AI for decades. We are seeing incresed adoption of AI with GenAi creating new opportunities, and making technology available for the non-business users.​ Many of you probably are already using or piloting with use cases such as knowledge management, recommendations, summarization, and of course contact center intelligence.
4. Tendu 1 minute (refer to Constanza will talk about this later) Topic Data not ready  But according to Gartner IT Symposium research, only 4% of participants reported their data is AI ready.  Note that this is a self-reported metric, so even this statistic may be inflated.      What happens when your data is not AI-ready?  Bias & hallucination  Inaccurate predictions  Excessive time spent with Data prep instead of spending time for model tuning or insights Gartner review by Manisha 4/23/2024 - Added 'Precisley View'Gartner review by Manisha Jain 2/13/2024 Hi Sue,   Thankyou for incorporating my suggested edits.   I have made further edits for you to make the deck compliant with the Gartner Content Compliance Policy. With my edits included and no other additional changes, this deck is good to go.   If you are using any Gartner references in the talk-track, please make sure that you don’t:   ·Cite any company-specific content from Gartner reports.   ·Promote your or your positive partner’s inclusion as an award or to imply endorsement by Gartner.   ·Use Gartner recognition or content as a competitive weapon.   There is no need to resubmit this request unless there are any further changes/edits.
5. Tendu 2 minutes​ Topic 3 challenges​ ​ Without AI-ready data, your business is exposed to a wealth of potential harmful impacts.​ ​ For example, without access to all the relevant and critical data your AI needs, you may deliver outcomes that are ageist, sexist, racist, classist, and more. ​ You may be familiar with the HR recruiting application that prioritized application from men. ​ Or the tutoring company that automatically rejected women 55 or older – resulting in a 6-figure fine. ​ Or the healthcare algorithm that failed to prioritize care for non-white patients. ​ Or even issue of common image generators producing images that lean into stereotypes regarding age, gender, race and more when producing images related to various jobs and roles​ ​ And without data of high quality that is well-governed, businesses are exposed to:​ Hallucinations that deliver incorrect analysis or recommendations – with high profile stories in the legal and medical spheres.​ As well as wholesale failures of AI applications – such as issues with autonomous cars if maps are not accurate to the pixel.​ And of course, at the top of many enterprise legal team’s lists – without proper governance, security and privacy of sensitive data I at risk.​ ​ Lastly, if data is accurate…. but lacking in context related to the real world, such as consumer insights, property attributes, and spatial relationships, irrelevant outcomes can be just as costly.​ You may have heard of Zillow’s $3.8B loss due to AI that improperly valued homes – due to lack of relevant property insights.​ Poor decision making based on missing context can have severe and costly impacts.​ And it can just simply miss the mark with users who see the lack of nuance and relevance in AI results​ ​ =============================​ ​ ​ Tackle Bias​ Bias in text-to-image generators biases flagged in July 2023:​ Ageism & sexism: Lack of lined faces on women, lack of non-binary representation, lack of representation of women in certain roles​ Racial bias: Lack of racial representation in workplaces and in specialized roles​ Classism, conservatism, urbanism, and anachronism​ ​ Minimize bias, improve accuracy and reliability, and enhance the understanding of the problem by ensuring fresh and relevant critical data, including unstructured data, is incorporated — on-premises and in the cloud.​ ​ Critical capabilities:​ Data integration​ Data governance​ ​ Ensure accuracy:​ Lack of data governance and data quality has led to high-profile failures & hallucinations:​ Inaccuracy of map data at the pixel level questioned in  autonomous vehicle collisions​ Lack of data completeness underlies chat hallucination in medical and legal use cases​ Poor governance of data and models has exposed organizations to leaks of internal data and biased outcomes​ ​ Increase accuracy by ensuring the data you're using is accurate, consistent, and fit for AI​ ​ Critical capabilities:​ Data governance​ Data quality ​ Data observability​ ​ The challenge: Lack of contextual relevance​ Without context into the nuances and dependencies of a given real-world scenario, the AI bases its inference or recommendations on only a small portion of the bigger picture. This can lead to incomplete, inaccurate, or contextually irrelevant results with potentially dangerous downstream impacts.​ ​ The solution: Spatial analysis and data enrichment​ For more accurate and contextualized AI outputs, you must enrich the data that fuels them with trusted third-party data and spatial insights.​ This proce
6. Tendu 1-2 minutes​ - (Tie your story to Enrich story while introducing Constanza)​ Identify & integrate all relevant data for training & inference​ Plan for data quality and governance of for AI models & pipelines​ Enrich data with 3rd party information and spatial insights to increase relevance​ ​ ​ ​ Do you have access to all critical and relevant ​ ​ ​ Less data for high quality is better than more data with poor data quality​ Conclusions or Takeaways for the title?​ Steps to successfully manage data integrity and drive better AI/ML outcomes​ Why location intelligence and data enrichment are critical for trusted business insights​ Strategize and drive your AI/ML initiatives with a business outcome driven approach ​ ​ ​ Taken together, these steps — data integration, data quality and governance, location intelligence, and data enrichment — ensure access to data with maximum accuracy, consistency, and context driving trusted business insights derived from AI/ML.​ OR: from Forbes​ The negative impact of bad data for AI/ML is exponential. ​ Delivering data with maximum accuracy, consistency and context is key for trusting the business insights derived from AI/ML.​ Without data integrity, we cannot trust the data and we cannot trust the business insights based on that data.
7. Topic: GRE is an innovator in ML applications CBC notes In 2019, GRE invested in the formation of a dedicated Analytics Team as part of the new Digital Transformation office. The team, comprised of international and experienced data science professionals, established GRE as the second European Real Estate Asset Manager to have a dedicated function for the development and management of proprietary machine learning applications. In an industry that was both the dream and the horror of any scientist: a greenfield, we started thinking on how we could support investments management to be more precise, informed and aware when taking decision on our portfolio.  We understood nearly 60% of change in value over a cycle ( 7 years) was not explainable using traditional real estate metrics, such as capital value and prime yield. 
8. Topic: This was the problem that the team was tasked with CBC notes In an industry that was both the dream and the horror of any scientist: a greenfield, we started thinking on how we could support investments management to be more precise, informed and aware when taking decision on our portfolio.  We understood nearly 60% of change in value over a cycle ( 7 years) was not explainable using traditional real estate metrics, such as capital value and prime yield.  So, we needed to find answers in the geographies and how and at which extent those geographies were bi-univocally influencing and be influenced by the Real Estate.  How demography, retail, services, investments, mobility, consumer spending, were shaping the face and value of neibourhoods, and eventually of entire communities. The beta version of City Forward was developed to answers those questions.
9. Topic: The result City Forward (Sue’s NOTE: “Transforming Real Estate through AI” is the tag line of City Forward? CBC notes Five years later, CF is a European leading location intelligence platform in the Real Estate but not only. With more that a dozen of use cases, and scalable applicability cross industry, and cross geographies we have reached unprecedent granularity, variety and forecasting power.  VIDEO? However, proved that the hypothesis was valid,  to scale such a concept in Europe, where characteristics of cities are so diverse, in a way that was at the same time, consistent, robust and scalable?
10. Topic: CBC addressing bias reduction We strategically incorporate a wide array of data, ensuring diversity in both source and type, crucial for minimizing bias. Our approach includes a comprehensive collection of European geographies, more than 11, covering northern, easter, wester, and southern Europe, enhancing representativeness and accuracy across diverse urban rural landscapes. Enabling cross cities learning.
11. Topic: CBC addressing data quality Internal data quality rules required Context, mobility, and social
12. Sue’s NOTE: Transforming Real Estate through AI is the tag line of City Forward (Back to Tendu) Topic: CBC addressing Data Enrichment CBC notes Data Enrichment for Enhanced Insights: By enriching AI datasets with detailed geo-addressing and location intelligence, City Forward unlocks deeper, more relevant insights. This enriched data layer transforms AI outcomes, driving actionable intelligence that is contextually aware and precisely targeted. AI is a topic of note. As a result of the combination of Satellite and OCR technology, City Forward has matured from a machine learning-based application to an artificial intelligence-based application. Without context, there is no such thing as artificial intelligence in the field of location intelligence, as we have discovered. If artificial neural network analysis is unable to extract crucial characteristics from the building and its surroundings in order to estimate energy consumption, then our CO2 emissions will, in fact, not be able to perform as expected. In order to provide significant insights, artificial intelligence (AI) is not merely a magic trick; it requires context and enrichment.
13. Tendu 1-2 minutes​​ Identify & integrate all relevant data for training & inference​ Plan for data quality and governance of for AI models & pipelines​ Enrich data with 3rd party information and spatial insights to increase relevance​ ​ ​ ​ Do you have access to all critical and relevant ​ ​ ​ Less data for high quality is better than more data with poor data quality​ Conclusions or Takeaways for the title?​ Steps to successfully manage data integrity and drive better AI/ML outcomes​ Why location intelligence and data enrichment are critical for trusted business insights​ Strategize and drive your AI/ML initiatives with a business outcome driven approach ​ ​ ​ Taken together, these steps — data integration, data quality and governance, location intelligence, and data enrichment — ensure access to data with maximum accuracy, consistency, and context driving trusted business insights derived from AI/ML.​ OR: from Forbes​ The negative impact of bad data for AI/ML is exponential. ​ Delivering data with maximum accuracy, consistency and context is key for trusting the business insights derived from AI/ML.​ Without data integrity, we cannot trust the data and we cannot trust the business insights based on that data.
14. Tendu 45 seconds​ Topic Journey​ While every data journey is unique, more often than not, it's a complex and ongoing process with many steps between defining a data strategy and deriving useful insights.​ The data journey never ends – it continues to evolve along with your business.​ ​ Josh recording.  Covers these slides in 2 minutes​  00_corporate_deck_josh_rogers_v1 (540p).mp4
15. Tendu 45 seconds​ Topic Precisely can help at every step​ ​ We believe you can more easily and effectively build Data Integrity when the core capabilities you need are working together. That’s why Precisely brings a complete set of differentiated capabilities to support your data initiatives now – and into the future. ​ ​ Our strategy services help you define, or refine, your data initiatives. While our complete portfolio of software and data products support every step along your data journey, so you can:​ Bring together the data your business generates – including complex IBM mainframe, IBM i, and SAP data in which Precisely has deep expertise ​ Catalog, cleanse, and validate your data so that your users can find, understand, and trust it​ Ensure proper governance of your data and management of your master data​ Enrich it with third-party datasets and spatial insights​ And activate it for optimal decision making and customer communication
16. Check out the eBook to uncover more about the foundational elements of trusted AI, the top challenges to trusted AI and how to overcome them, and more