Edgar Baum

Edgar Baum

In this article, Toronto-based LES Member, Edgar Baum, shares his insights from the recent LES Silicon Valley all-day, AI-themed conference, held at the Oracle Conference Center in Redwood City, CA. Edgar is a global leader in financially measuring market demand amongst consumers and businesses to inform strategy, forecasting and valuation.  Most notably, Edgar has participated in the drafting of the LES Intellectual Capital in the Boardroom Standard. He has worked with executive leadership and boards at Cirque du Soleil, Intuit, P&G, and Walmart amongst others on strategic measurement, business valuation and customer acquisition strategies. Earlier in his career, Edgar built Procter & Gamble’s customer profitability model across over 30 brands and 8,000 SKUs. He is in his 10th year as a Graduate Lecturer on the Finance of Brand Management at the University of Toronto

I had the pleasure and honour of participating in Licensing Executives Society (U.S.A. and Canada), Inc. Silicon Valley Chapter’s first in-person annual conference in five years and it was a good one!  As an LES member who flew in from across the continent, it was well worth my time, and I hope my summary does the speakers and content justice. A big thank you to Oracle for hosting the event at your welcoming campus.

The overarching theme of this conference was AI Unleashed: Unicorn or Trojan Horse.  The lineup of speakers and panelists ranged from innovative startups that are less than two years old Mark Leonard to contributions from one of the world’s most valuable companies care of Shaloo Garg at Microsoft .

This conference further reinforced the need for broader recognition and adoption of the Intellectual Capital in the Boardroom Standard that I’ve helped co-author at LES since 2016 with Paul Roberts, Ron Laurie, Peter Harter and others – see abstract.  This Standard systematizes how the board and leadership of a team can build business strategy while being conscious of opportunities and risks that cannot be protected or mitigated by IP protections ‘today’ but are business critical and can be protected ‘tomorrow’.

TL/DR: AI is in the Wild West of legal protections and commercial viability.  We don’t even have a track record yet of how effective AI is at reliably generating value.  Regardless of collateral damage along the way, AI is here to stay and upend most of the existing human and business structures we have in place.  AI will force a pace of innovation humanity is uncomfortable with and require inventions across a range of industries to support its effectiveness.

PANEL 1: Deus ex Machina: AI – Catalyst of Change or Chaos moderated by LES-SV Chapter Chair, Efrat Kasznik

  1. We can’t forget that a key limit of AI-only adoption is user trust – very few people trust a ‘0 to 100’ AI but they’ll trust it if there is an opportunity for human verification or augmentation along the way. Most legacy companies, especially healthcare and pharma, don’t have easy license to enter into AI organically – in fact it is easier for them to partner or acquire – thank you Maital (Shemesh) Rasmussen
  2. We’ve evolved from Cloud-native companies to AI-native companies built on generative AI and these valuations have a massive premium and it is still hard to judge how many will succeed. A big question to ask in investing into AI is whether there is a scalable use case for the Get AI solution 72% of start-ups fail – the number one success factor in the remaining is a constant inflow of revenue People like futuristic solutions but money is made in the pragmatic ones. Big Data has been replaced with Data at Scale for AI – thank you Shaloo Garg
  3. AI is forcing new areas of patent, copyright, and other legal protections, many of which haven’t been figured out yet – I.e. who owns derivative work?
  4. Understanding AI protections in a nutshell: Using AI to make a better version of what exists already as a human process is not easy to protect.
  5. Creating a new novel way to have AI models work- that one is able to patent.
  6. Have an AI create a solution: cannot patent (Except in South Africa but no one else recognizes it globally)
  7. Co-developing new solutions or products with AI tools: patentable – thank you Jordan Becker for the above 3-7 points.
  8. BIG QUESTION: (this one was mine) Who ‘owns’ the output/or confirmation of the outcome data from AI tools? We can’t figure out how these AI models got to a summary or interpretation of initial data (there is no audit trail in most modern AI systems).  Essentially, who ‘owns’ the results of the outcomes from execution recommended or acted upon by the AI? Is it the data owners or the AI model builders, both or neither? i.e. if you gave an AI tool bad data and it made what ultimately resulted in a bad outcome, who’s fault is it. If you gave it good data and it resulted in a good outcome, who gets to claim credit?
  9. REPLY: Hearing from the panel and others, this is a major, unanswered question and ties into who owns underlying data and the role of legal protections.  In short, watch this space!

PANEL 2:  AI: Revenge of the Hardware moderated by LES-SV Chapter Board Member, Michael Pierantozzi

  1. AI software will only go as far as AI hardware and energy usage will take it – the next stage of AI needs to be energy efficient meaning smaller, more focused models.
  2. Developing hardware for AI needs to be partner-based and ecosystem-based and the biggest constraints are power consumption, distance between the data storage, the computing environment, and the recipient of the request/output hence the role of telecom companies is important.
  3. 460 Terawatt hours used by AI & Crypto in 2023 – equivalent of 50 Nuclear Power Plants – this will need to double by 2026 to keep the growth curve going.
  4. For AI to remain scalable, it will need to migrate to Edge AI i.e. automotive AI is Edge AI (AI functionality on the device rather than on a central server – this requires completely different technology solutions) – thank you David George , Louis R. and Abul Nuruzzaman for the above points made

KEYNOTE: The Future of Generative AI from Greg Pavlik (Senior Vice President, Oracle Cloud Infrastructure)

  1. Four quarters out is about as comfortable as Greg at Oracle is comfortable in attempting to forecast.
  2. There is a belief in industry that once working well, Gen AI will have a greater impact than the internet, some think even more than agricultural revolution.
  3. The greatest impact is on the knowledge worker who has to summarize information, conduct structured synthesis, combine contextual data i.e. making a travel plan, sending a relatively similar email.
  4. In brief, AI today is solving for generation, extraction, and summarization with evolutions in accuracy and multi-modal content integration as the next step. The next major milestone that we can comfortably foresee is for AI tools to reason and become assistants, reliably automate, and become independent in tasks that don’t depend on human intervention but don’t impair humans in the process at the same time. In the near term, this can show up as agents for retrieval augmentation generation, NL2SQL, Code assist, travel booking, and Sales/transactions. Thanks Greg!

PANEL 3: AI in Action: Paths to Commercial Success moderated by LES-SV Chapter Board Member, Pallavi Shah

  1. There are a lot of AI solutions looking for a problem mudding the waters. Not all AI is the same, the best tools are applying solutions to behaviours that are already there – thank you Zack Cable
  2. AI enables the collection and visualization of data, a process that historically was prohibitive in cost and time. Modern strategy is enhanced by bringing knowledge into visualization to inform new insights – thank you Jiyoung Choi
  3. CFOs need to be very conscious of the role of patents and IP at driving value in the organization – thank you Subash Krishnankutty
  4. There is no such thing as IP strategy separate and apart from business strategy. The detectability of patents is most important to enforce them – this needs to be taken into consideration when considering patents as being protective of the business model. It is also crucial to know with AI when am I the product vs the customer – thank you Tom Mavrakakis

PANEL 4: VC Perspective on the Evolution of AI and Systems Applications moderated by LES-SV Chapter Board Member, Darius Sankey

  1. We need to be thinking out 15 years+ when it comes to infrastructure and capacity investments. We’re at the infancy stage of understanding how to balance between centralized and edge AI – much of this is a hardware problem and connectivity problem as 5G matures and transitions to 6G. Europe is ahead in thinking of and solving for this problem, well in advance of the US today. No single entity can solve for this, it will need to be a consortium of academic, government, and corporate based on royalty and licensing models – we need pure research, regulation changes, and entrepreneurship – thank you Darius Sankey
  2. Super Unicorns are under even greater strain and dependency on hardware than smaller AI start-ups because the gross value of their upside is on the hardware being there for them in the future, ready and functional. The lifecycle and payback of any single solution or package is too hard to predict, need many initiatives at the same time – thank you Hermann Schindler, Ph.D.
  3. It is critical to try and map the value chain over time – how much will current funding cover? What type of funding will be required in the future?  Total addressable market (TAM) is a very misleading estimate because a single solution doesn’t cover all of it – need to carve out what and when portion of that TAM will be solved for by any hardware technology. We’re in the midst of a hardware revolution right now that is not recognized and talked about enough.  Whole new technologies are required for Edge, cloud, data centre dependent on energy efficiency, communication speeds. These system models can be commercialized by getting existing players to opt-in and commit to royalties and licensing that accelerates the investment use case – thank you Reza Khaj
  4. BIG QUESTION from Efrat Kasznik: The VC business model is best suited for smaller investments, giving VCs a minority share in companies that can get to exit and provide the largest return on investment in the shortest period of time; that is why software/ SaaS has been the most common segment for VC investment.  Are VCs even suited for the large amounts of money needed to be invested in an AI environment, and the long periods of R&D that it would take to get to commercialization?
  5. REPLY: Big companies have done the easy stuff; small companies need to identify systems solutions and solve for those. Not all growth will come from new revenue, but rather, clear cost savings for existing practices – thanks Darius Sankey
  6. REPLY: Hardware is hard to scale fast, Nvidia was 20+ years in the making for the past 3 years of success – it was not predictable 20 years ago – thanks Hermann Schindler, Ph.D.
  7. MY TAKE: AI is so nascent that we don’t even have an AI marketplace.  AI hardware investment looks more like a Hollywood production or, centuries ago, trade ships!

FIRESIDE CHATwith ScaleIP CEO, Mark Leonard and ScaleIP Investor, Eric Ver Ploeg; moderated by LES SV Chapter Board Member, Kent Richardson

  1. ScaleIP is an example of a company that is at the intersection of Machine Learning and AI in being able to match patents/technology to companies that can commercialize them. Builds on the themes of the day on how AI can facilitate commerce that wasn’t possible before.
  2. As per Kent Richardson, this approach unlocks the previous one which was intermediated by humans and what they knew consciously – this enables a combination of research and insights that takes astronomical costs of data gathering and synthesis and makes them affordable.
  3. For Eric Ver Ploeg as an investor, this is the type of company he truly likes, one where use of AI enables new category creation with a ready demand but existing solutions/alternatives are inefficient or expensive – the unit economics make sense for all parties.

My Biggest Takeaways:

  1. The conference reinforced that we’re still very early in Geoffrey Moore’s curve of Innovators > Early Adopters > Early Majority > Late Majority > Laggards.  Considering that analytics have penetrated the depths of laggard organizations, AI as the next wave makes a lot of sense but it is not as far along as many in AI, and in the room, would like to admit to – too many of us in the room are the innovators and early adopters ourselves. We could collectively be wrong much more frequently than right about AI for the foreseeable future.
  2. Data Quality is still the big elephant in the room and not enough companies are acknowledging whether AI customers actually have good enough data to take advantage of AI systems.
  3. AI is at it its infancy, circa iPhone apps in 2007.  The success of AI is dependent on contextual efficiency and low cost of utilization – hence, eventually, a company will run numerous AI apps to take advantage of contextual needs for HR, Finance, Sales, etc.
Share this article: