Dr. Yuvraj Kumar – Chief AI Officer
Whether your organization needs a chief artificial intelligence (AI) officer is a topic where there have been differences of opinions. However, the primary idea is to have someone who heads or leads the AI initiatives across the organization, who is a master of AI, possibly a Ph.D in the field of Artificial Intelligence, not a data scientist, or machine learning engineer, neither an evangelist of Artificial Intelligence, but actually who is an expert of Artificial Intelligence.
The designation could be Chief AI officer, Vice-president (VP) – AI research, Chief Analytics Officer, Chief Data Officer, AI COE Head or maybe, Chief Data Scientist etc. One must understand that building AI/machine learning models and deploying them in production is just one part of the whole story. Aspects related to AI governance (ethical AI), automation of AI/ML pipeline, Machine Advocacy, infrastructure management vis-a-vis usage of cloud services, unique project implementation methodologies etc., become of prime importance once you are done with the hiring of data scientists for building the models. This is where one would need a person who leads the different AI initiatives such as the one mentioned above, hence the AI ML Machine Advocacy Council came up with all the strategy to position itself in the industry by entrusting Dr. Yuvraj Kumar with all the responsibility and accountability of representing AIMMAC worldwide as the world’s first and industry’s first Chief Artificial Intelligence Officer (CAIO).
Key Responsibilities of the AIMMAC Chief AI Officer:
- AI Roadmap vis-a-vis Business vision/goals/initiatives (VGIs)
- Ideas & insights into newer AI-powered business models
- Projects/products implementation methodologies/processes (agile etc.)
- AI algorithms, research & related roadmap
- Technologies & related roadmap
- AI platform/products architecture/design/implementation vis-a-vis cloud AI services
- AI governance (ethical AI) and related processes
- AI infrastructure vis-a-vis cloud AI services
- AI automation (pipeline) projects
- AI/ML models quality assurance strategy
- AI/ML models continuous delivery/deployment strategies
- Communication with customers/partners/media/internal
- Hiring/Training/Mentoring
- AL / ML Global Advocacy
In Brief – Responsibilities of Chief AI Officer of AIMMAC:
- Identify AI projects of high business impact including both product-related projects, and also research projects. Collaborate/communicate effectively with product owners representing different product lines.
- Prepare an annual roadmap for implementation of different AI projects.
- Provide ideas and insights into new business models that could be enabled using AI. Become a go-to-guy for C-Suite executives for implementing AI in their functional areas.
- Layout the AI projects/products implementation processes (ML model development lifecycle) including project inception, exploration phase, model building phase, model deployment, and model retraining. Ensure the project implementation governance on the ongoing basis.
- Prepare the plan and oversee the implementation of the AI platform which will be used to deploy and host ML models.
- Play a key role in deciding technologies and related roadmap for development, testing, deployment of AI/machine learning models.
- Prepare the plan and oversee the implementation of quality assurance processes in relation to model testing by different stakeholders; Model performance testing, model acceptance testing by product managers/consultants/customers, other forms of testing as applicable (such as metamorphic testing, dual coding testing, blackbox, white-box testing etc)
- Prepare the plan and oversee the implementation of continuous delivery/deployment strategies (A/B testing, canary testing etc) of machine learning models.
- Prepare the strategy/plan for ethical AI and oversee its implementation across different AI projects. Get involved with the interaction related to ethical AI with stakeholders including customers/partners AI governance team, auditors, regulators on the ongoing basis.
- Prepare the plan and oversee the implementation of machine learning pipeline automation to be used for automated ML model retraining/testing across different AI projects.
- Prepare the plan and oversee the implementation of AI infrastructure to be used for model training/retraining and model deployment in production. Ensures the use of cloud services vis-a-vis local infrastructure for fulfilling different requirements.
- Hire a team of data scientists; Keep a check on newer hiring requirements on the ongoing basis. Provide training and mentoring to the team
- Take part in communication with customers, partners, media stakeholders
Core Skillsets being represented by the Chief AI Officer of AIMMAC:
- Up-to-date with AI research areas and current trends
- Nuances related to building machine learning models including data preparation techniques, feature engineering techniques, building machine learning (ML) models and related ML algorithms
- Stay up-to-date with current developments in the field of AI – Safe AI, Ethical AI, Fair AI etc.
- Cloud AI/ML services on AWS, GCP, Azure
- Cloud computing services
- Law and Advocacy for Machines
- Knowledgeable about Big Data technologies and related cloud services
- Software development lifecycle, project implementation methodologies such as agile etc
- Software engineering principles
- DevOps concepts for ML pipeline automation
- Infrastructure knowledge for deciding physical/virtual m/c for model retraining etc.
7 Reasons – Why you need a Chief AI officer (CAIO):
AI’s rapid surge in value-add has left companies struggling to adopt this highly complex technology. Executives struggle to understand it. At a basic level, the terminology is confusing, machine learning, deep learning, reinforcement learning, AI, etc… The business use cases are unclear, and the experts are mostly in academia, have their own startups or are at top tech companies.
Enter the Chief AI Officer (CAIO).
How businesses try to adopt AI.
The adoption process goes something like this: An executive reads or is told multiple times about how AI can do X for their business. The CTO or CIO looks into it and concludes that AI can probably help the company save costs. However, the benefits, AI approach and possible downsides may still be unclear.
Next, the company might decide to hire a Ph.D. or two with relevant research experience. They’re directed to build X system and are left alone to work it out. Eventually, the results and expectations don’t match, and the team is disbanded or re-focused on data science applications. Soon, the business buckets AI in the “hype” category and moves on.
We get it, getting AI to work is hard – especially under business constraints.
1. Your CTO or CIO is an expert in engineering – not AI.
Great CTOs know how to monetize software. They know the best ways to cut costs or how to approach problems using the latest software engineering paradigms. However, he’s / she’s unlikely to know about the latest trends in AI and where they could help the company.
The field of AI is huge. Academic paper submissions to Arxiv “continued on a very fast growth rate 2011-2015, and on an even faster growth rate since 2015 (driven largely by the computer vision, machine learning, and computational linguistics communities).”
https://arxiv.org/help/stats/2017_by_area/index
To keep up with the latest trends, AI researchers read papers on a daily basis, attend conferences, and host private research presentations from visiting scholars. Just in the past few years, the amount of new research published in machine learning has grown faster than any other field. Although many papers are small improvements, your organization needs an expert to sort out the key developments and the implication to the business. Could be as simple as a new approach to recognizing text that could suddenly open up a whole new business line for your company.
The CAIO should be someone who is deeply knowledgeable about AI and familiar with current approaches such as deep learning, reinforcement learning, graphical models, variational inference, etc. Without this expertise, they might deploy approve approaches which are slow to implement, costly to maintain, or that don’t scale.
2. A CAIO is your lifeline into the latest in academic research.
It’s no secret that the world’s top AI researchers at big companies like Google and Facebook also hold academic links.
- Yann Lecun, the head of Facebook AI, a deep learning pioneer, also heads NYU’s Computational Intelligence, Learning, Vision, and Robotics (CILVR) group.
- Fei Fei Li, head of Stanford AI Lab (SAIL), also heads up Google Cloud AI.
- This year, Facebook raided CMU’s AI lab with many professors and students joining part-time.
- In Canada, Yoshua Bengio, another pioneer of deep learning, remains a full professor heading up Montreal Institute for Learning Algorithms (MILA) but has managed to bring Google Deepmind, Google Brain, Facebook AI and a few other labs to Montreal to work with his students.
Top tech companies have found this loophole which gives them direct access to the world’s top AI graduate students. A strong academic link also allows for partnerships with these labs which can help tackle hard business problems in exchange for the ability to publish results.
To attract top AI talent, hire top AI researchers. To retain talent, your AI team must be allowed to contribute to the open-source AI community and publish papers. If you don’t, they’ll go to Google or Facebook where they’ll have that freedom!
3. The C-suite needs a trusted expert who can deploy AI to create new business lines.
Many organizations fail to fully leverage AI because the C-suite often doesn’t understand AI capabilities. Hire an expert who understands the technology and understands how to solve business problems with it. An AI expert in the room will abate concerns about the revenue impact of a new AI system and potential business risks.
We’ve seen colleagues with promising ideas get shut down by organization leaders because they don’t understand the impact this new system can have on the business. Don’t let the c-suite’s lack of expertise in this field keep your organization from making big AI-driven changes with huge potential upsides to your business. It’s like not using the internet because you don’t know how TCP sockets work.
Andrew Ng, the founder of Coursera, adjunct professor at Stanford University and a top AI expert, argues that the Chief AI Officer is someone with the “business expertise to take this new shiny technology and contextualize it for your business.” In essence, someone with both the strong academic background and the business acumen to solve business problems using AI.
4. A good CAIO adds perspective to the C-suite.
Don’t launch your next product or business line without thinking about how AI can help. AI’s use in business is so new, that it’s unlikely anyone in the executive suite is thinking about new business lines that are now available because of AI. Look for problems that are hard to scale, or that requires a set of complex rules, these are prime candidates for AI.
Beyond the technical capabilities, your CAIO needs to have a good sense of the business. This is a person that knows when NOT to use AI. A good CAIO will make sure their team isn’t looking for places to apply AI, but instead looking for problems that could benefit from it.
5. Data is a revenue stream.
By now, companies are aware that their data are hugely valuable. If you believe this premise, then it follows that you’re likely leaving a lot of money on the table by sticking with old methods that are known to underperform other algorithms. A classifier that can segment users 20% more accurately means that you’re likely to put the right products in front of that user. Why settle for the machine learning models that give you 80% accuracy when more modern systems can get you to 90%?
The CAIO marries the analytical and business skills required to supercharge your data monetization strategy.
6. Signaling.
If your business wants to signal that you take AI seriously, then hire a CAIO. AI is an afterthought in most organizations. Don’t make the same mistake. Signaling will help you attract top talent, rebrand your company’s public appearance and signal to investors that you’re still innovating.
7. Ethics.
AI’s use for different cases has come under scrutiny in past years. Recently, a revolt within Google pushed the company to pledge not to build AI weapons. The AI research community has started to voice concerns over ethics in the past year. As a result, top researchers are unlikely to apply AI to problems they deem unethical. The CAIO can serve as the voice for an organization and drive the use of its AI towards profitable, yet ethical use cases.
The ethics aren’t just about what the AI is applied on. It’s also about the data handling which can introduce biases. Having a CAIO will signal to the world that your company takes AI ethics seriously. Large companies don’t want to get caught in a PR problem around AI if they can help it.