Some insights from "Building Enterprise AI Excellence"​, from North America AI & Big Data Expo, October 2022

Some insights from "Building Enterprise AI Excellence"​, from North America AI & Big Data Expo, October 2022

Monday, October 10, 2022

As I mentioned recently, early October in Santa Clara, CA at the AI & Big Data Expo World Series, I had the opportunity to capture interesting insights from SMEs and business leaders around #ai adoption and AI challenges. Find here some takeaways I want to share about "Building Enterprise AI Excellence" (more capsules to come about other topics and presenters during the #aiandbigdataexpo). My take on this intervention by Daniel Wu (head of AI & ML Commercial Banking at #jpmorganchase) is how relevant it is for business leaders to gain a perspective on AI/ML adoption with a holistic lens. Everyone working in high-tech (and also other verticals!) has been exposed somehow to the positive benefits that emergent tech like AI can bring thanks to extracting value from the increasing #computing power we now have, solutions automation, making #unstructureddata useful, faster #decisionmaking, and so on (too many benefits to mention here!). However, adopting new tech as AI has many aspects (beyond the business problem it solves). To be really prepared to move forward with #aiadoption, any organization needs to consider AI as a total new aspect of the business, encompassing the elements described in the framework presented here. According to Daniel Wu, every year enterprises are investing more money in adopting AI (as expected). However, a proper framework and solutions to face challenges brought by such adoption, are needed to prepare organizations in this transformation. The Framework to drive this process, should consider five different angles: Data - considering its quality, cleansing process, accessibility and #datasilos within organizations. Compute - Managing security, #privacycompliance, cloud migration, and carbon footprint, among other challenges. Talent - Navigating through talent shortage, lack of #diversity and corporate AI literacy. Operation - supporting scalable models with dedicated teams to manage AI-related processes. Governance (this one being my favorite!) - where lack of ethical knowledge and #regulations, as well as absence of AI boards withing orgs, are creating exponential risks for consumers and companies adopting AI-based solutions.

AI Adoption in Enterprise
  • In 2021, around $93B were invested in AI WWide by private organizations (twice as much as previous year). USA consolidated 56% of that investment, followed by China with 18% and EU with 6%.
  • AI adoption by industry: High Tech and Telcos lead the way, followed by H.Care and Financial Services.
  • At adopting AI, organizations start to perceive same level of risk as they do on cybersecurity threats (McKinsey 2021).
  • 48% of respondents think that AI use should be regulated more strictly– (IPSOS World Economic Forum).

AI Adoption Challenges

1.      DATA is broadly siloed withing orgs. Adoption of AI ends up being affected when information is not adequately available internally to make AI- related processes robust.

2.      COMPUTE: Organizations moving from private to public (or hybrid) #cloud need to overcome privacy challenges related to #datasovereignty , #compliancemanagement, and #security. Not to mention the exponential computing power that requires more controls making it harder for organizations to catch up.

  3.      TALENT: Lack of diversity (which contributes to #biased models) is still a concern. Percentage of women into AI hasn’t changed in the past 10 years plus, AI professionals are clustered in very specific points in the world. Notice how concentration of AI professionals in the top 5 cities -WWide- are San Francisco, New York, Boston, Seattle and Bangalore.

Talent Challenges in AI Adoption

4.      OPERATION - A major need of moving from model centric approaches to data centric ones, in order to overcome key challenges as model lineage and reproducibility.

  5.      GOVERNANCE: The trade-off between Time-to-market vs do-it-right opens ethical concerns when launching new AI-based programs. Additionally, lack of #responsibleai (or integration of it) into the decision-making process, and scarce regulation and laws, generate the need for organizations to dedicate more resources to extend the governance arm, either through third-party AI solutions or more internal resources.

The game is on, and as organizations rush into adopting AI with the north star of being innovative, adaptable, improving time to market, beating competitors, serving customers better, etc, at the same pace, it is necessary for corporations to gain awareness and trigger actions on how to manage that "black-box" called #machinelearning model and #aiforbusiness and the outcomes associated. Such outcomes are impacting people, brands, and organizations, and it is in our hands to make that impact a good one.

Cheers,

Mauricio / AI for U founder

No comments yet
Search