AI receives an abundance of media coverage on an almost daily basis and is a top strategic priority among industry business leaders and world governments. The question on the tip of everyone’s tongue is: “Where is AI taking us?” Will AI usher in a new golden period of prosperity or will it take us to an increasingly dark place that would make Ready Player One appear sunny by comparison? While pundits offer many predictions, new technologies tend to confound forecasts.
With the prominence and pervasiveness of AI chatter, it is easy to forget that this latest AI cycle (already one of the longest periods of sustained interest in AI in history) is a very recent phenomenon – we are barely getting started. While we are fast learning about this new power, we still know only so much. Nonetheless, in some quarters there is a growing sentiment slanted towards a dour scenario, where AI inordinately benefits large corporations, eating jobs and exacerbating wealth concentration. For example, the IMF has expressed concerns about how AI could worsen wealth inequality among households, McKinsey has written about how AI could widen the US racial wealth gap by $43bn per year, and The World Economic Forum has expressed concern about how AI will widen economic disparities between the Global North and the Global South.
At Core, we take a different view: we believe that AI will be a new force that will create massive economic opportunities for people as “producers” and earners. In doing so, AI will help to resolve the economic imbalances that have persisted for decades by empowering people individually and collectively. Below are some early signals that inform our alternative AI vision, one that we believe is not only plausible, but in fact likely.
The Green Shoots: Increased Access to AI Compute, Growing Data Availability and New Models of Data Ownership, Public discussion of AI governance
Democratization of Access to AI Compute: We see new platforms that provide a Plain English interface to AI compute as taking down the barriers to access for AI (in the future you will not need to be an engineer to take advantage of AI for programming). These Plain English interfaces, complemented by the introduction of Facebook’s Llama (a viable and competitive offering to ChatGPT for certain applications) as a free service opens up the market to entrepreneurs and developers who cannot afford ChatGPT or expensive AI engineers. In addition, NVIDIA’s “DGX in the Cloud” is increasing the availability of AI to SMBs and the mass market of developers further reducing AI cost and implementation hurdles. Using “DGX in the Cloud” companies and individuals can rent capacity on DGX servers (specialized Nvidia servers that specialize in using GPUs to accelerate deep learning) rather than needing to purchase one of Nvidia’s expensive (and constantly sold out) DGX supercomputers.
We believe that making AI available broadly to innovators is critical to counterbalance the tendency of large companies to skew investments in new technologies, such as AI, to drive greater efficiency in “rent-taking” activities versus investing in productive new initiatives that aim to open new markets and enlarge the economic pie.
Data Availability and Ownership: Open data policies (including Open Banking) and new privacy rules will provide greater access to data, the essential fuel for AI solutions. We also foresee the emergence of new data commons within communities of interest that will make critical training data broadly available to innovators. As one example, the Common Corpus initiative by HuggingFace makes available to developers a substantial publicly available dataset (without using copyright content) that can be used for training LLM models.
Regulation and self-regulation: We also note the proactive approach that governments are taking toward creating a regulatory framework for AI to help ensure a positive path of development. In contrast to the lagging efforts to regulate consumer data ownership and usage, where we are still limping toward a national framework for governing data ownership (and where we are optimistic that GDPR and CCPA will eventually pave the way for an overarching federal framework of data governance) the efforts to create a balanced approach to AI regulation appears decisively more proactive. For example, the Biden Administration issued an Executive Order on the Safe, Secure and Trustworthy Development of AI and earlier this year the European Parliament formally adopted the EU Artificial Intelligence Act.
While many may regard regulation as a potential negative to innovation, we believe that setting the ground rules clearly and promptly will facilitate and accelerate the formation of AI startups. With material regulation appearing inevitable, the persistence of ambiguity on the rules of the road creates an overhang for less well funded startups (which may not be able to afford the cost of potential course correcting when final regulations emerge) while doing little to slow down the activities and investment rate of large companies. Resolving this overhang sooner rather than later will be a helpful factor in distributing the AI innovation opportunity more broadly.
In addition, new models of governance, such as Constitutional AI are emerging. Constitutional AI is being championed by one of the largest AI software companies, Anthropic, which is seeking to build a platform that opens the black box of deep learning models. We are optimistic that Constitutional AI will trend toward governance frameworks that can be established by humans and audited by humans to provide greater control over deep learning platforms and increase confidence in driving toward less biased and more equitable outcomes.
We highlight the above green shoots while acknowledging that powerful new technologies have both a bright side and a shadow side, that duality comes with the territory of tech innovation (think gene editing, nuclear engineering, and social networking). How things will play out, well it’s in part up to us to determine. We believe that with the proper guardrails in place (social standards, self-regulation and government regulation) the evolutionary arc of AI can be skewed toward the greater good.
We are big believers that when it comes to technology-led innovation that necessity is the mother of invention. It seems to occur with regularity that new technologies almost magically arrive on cue at pivotal times to help humans to address seemingly intractable challenges. Whether this deus ex machina phenomenon is the result of human ingenuity being consciously channeled toward that outcome or larger forces acting upon us beyond our current understanding, as per the laws of Thermodynamics, all imbalances (or “gradients”) eventually resolve. We have some huge wealth imbalances that have been building for multiple decades – and the pressure is ready to be released.
Where Core is focused: Our AI investment themes
AI solutions will magnify incomes of SMBs and independent workers by empowering people as “producers” (not consumers!). At Core we believe that AI will massively amplify individuals, empowering them as “producers” to build new companies and pursue 1099 side hustle business opportunities. While the biggest tech will sustain, mid-size tech and most traditional companies will lose vitality as AI-driven cost-cutting will continue for years to come. Liberated labor will discover a new vocation as “producers” and build lean models to serve people across every sector (who really wants to make a living clipping universal income coupons?).
AI will also reduce the cost of essential goods and services (e.g., healthcare, education, and food) for building healthy households AI’s impact on these goods and services will be deflationary. It will drive efficiencies such that less financial and human capital inputs will be required to deliver these essential products and services. More broadly in the US economy, we will see a break with recent lackluster productivity growth trends and see a return to the rapid tech-driven productivity growth of 1995 – 2004. We may be already seeing the early signs of AI filtering into the government labor productivity numbers. The Richmond Fed recently wrote about how “labor productivity is popping” (in response to three strong quarters of productivity growth exiting 2023) and commented: “labor productivity growth is up 2.7 percent year over year [in 2023]… This growth in productivity has been robust compared to recent history.”
Using AI, households achieve mastery over their finances, health, residences and all the obligations, assets and risks that life entails. Drawn to the potential benefits of AI, people will increasingly embrace autonomy in key domains of their lives (see the blog post on The Six Levels of Autonomous Finance by my colleague, David Roos). Just as in the 1930s and 1940s, rural electrification and the adoption of white goods (washing machines, refrigerators, stovetops) liberated households from the drudgery of manual housework, so will AI liberate people from the cognitive load of managing financial and health risk and complexity. In addition, households will be advised increasingly by uncompromised AI fiduciaries who will replace humans and offer their services at a fraction of the cost, massively increasing access to advice and fiduciary services for lower-income individuals. So often in financial services, intermediaries, claiming to offer services in the best interests of individuals, including intermediaries acting as fiduciaries, have incentive systems in place that skew or even corrupt their advice to the disadvantage of the small guy. In contrast, AI will be our reliable and objective financial advisor, complementing or even replacing financial services professionals and fiduciaries who are susceptible to “incentive-caused bias.”
New data commons emerge. Greater access to data will be critical for creating broad-based opportunity in AI. New data trusts (some for-profit, others non-profits or community projects) will emerge for sharing information broadly within communities of interest and they will do so at a reduced cost. We have seen the rise of these models in the past. For example, Insurance Services Office (“ISO”) which was founded in 1971 as a non-profit (and eventually purchased by Verisk) to serve the mid and long-tail property and casualty insurance industry with a “give to get” model for sharing valuable claims and underwriting data. While large insurers had sufficient internal data to appropriately price their insurance policies, smaller insurers did not. By subscribing to ISO smaller insurers could obtain access to industry datasets at scale to inform their product pricing and underwriting decisions so as to better compete with large insurance carriers. We are keeping an eye out for models that are similar to ISO that can support AI innovation within industry segments, particularly financial services, and help level the data playing field between large and mid-to-small size enterprises.
Small is beautiful: AI propels the “lean startup” to a whole new level. Increasingly startups will challenge incumbents in massive markets and do so with hyper efficiency (i.e., limited human/financial capital) while others will deliver impressive profits while pursuing TAMs < $1bn. With the leveling of access to AI Compute and to training data the insight that derives from closeness to the customer and the ability to establish trust with the customer (areas where startups tend to excel) will amplify in their importance as a source of competitive advantage, providing a counterbalance to large company operational scale, proprietary data wealth, and large GPU budgets.
AI amplification of small startup teams will present unanticipated threats to larger traditional businesses across a variety of sectors (Perhaps a small team of developers will replicate and disrupt Salesforce.com. Might we find it possible to crowdsource the design and production of a jet-liner?). We foresee an SMB boom and the emergence of a US-born Mittelstand of AI-powered industrial and services companies. Along those lines, this Forbes article talks about how AI can reduce the cost and expertise needed to deploy digital manufacturing technologies and automation / robotics and help to narrow the competitive gap between small and large manufacturers.
Realizing a Vision for the Future: AI-powered democratization of prosperity
The AI revolution will be more than a 10-year technology cycle, it may continue for decades. The Internet’s original promise of decentralization remains intact and AI (complemented by Web 3.0) will reinvigorate its course. This game is only getting started: we are still at the first at-bat of the first inning, and this game will likely go into extra innings.
The impact of new tech can be so comprehensive that its long-term effects can be difficult to anticipate and plan for. However, our collective aspirations and belief in a better future will play a critical role in determining how this new power is channeled, and we remain optimistic that the outcome of the current public debate and discussion will direct AI in a positive direction.
Reflecting on this moment in AI development, we recall President Kennedy’s “moon speech” at Rice University. Replacing the word “space” with “AI” in JFK’s comments highlighted below conveys our feelings about the magnitude of this moment in time for AI innovation.
“We have vowed that we shall not see [AI] filled with weapons of mass destruction, but with instruments of knowledge and understanding… Our hopes for peace and security, our obligations to ourselves as well as others, all require us to make this effort, to solve these mysteries, to solve them for the good of all… For [AI] science, like nuclear science and all technology, has no conscience of its own. Whether it will become a force for good or ill depends on [us]”
Coming full circle we return to the question: “Where is AI taking us?” Core’s response is that AI will bring about a period of mass financial freedom and economic opportunity. We are seeking entrepreneurs who share this vision of change, which we refer to as “AI for AI(I)”, and are eager to partner with them. If you are building a company that seeks to democratize the benefits of AI, we’d love to get in touch and explore how together we can make that future happen.
