Conference Presentation, Panel
The Rising AI Economy: A New Era in Tech | Milken Institute Global Conference 2024
Milken InstituteNicole Valentine, Seth Boro, Les Borsai, Sachin Dev Duggal, X Eyeé, Martin Fichtner, Sachin Chakrabarti
- Sachin Chakrabarti predicts a rapid technological transformation mirroring the dial-up to broadband shift, stating that significant changes to the known landscape will occur within the next 12 to 24 months.
- Anticipating continued GPU usage rates, Chakrabarti forecasts that AI chip power consumption will equal the total power consumption of Europe within approximately 16 to 18 months.
- Open-source models running on edge devices are expected to enable the creation of remarkably realistic synthetic content, with risks of widespread damage emerging before mitigation strategies are implemented.
- Seth Sternberg forecasts that Generative AI will deliver exceptional market value and increase corporate relevance over time, despite current hype cycles, though governance and data ownership discussions will remain in early stages for the foreseeable future.
- Martin Lau projects that over the next 10 to 20 years, every business will have an opportunity to utilize AI for efficiency and margin expansion, eventually leading to professions such as medicine, financial advising, and individualized teaching being entirely AI-driven.
- Chakrabarti warns that companies relying on high revenue with "wrappers around an LLM" may face investor skepticism, while established firms with systematic AI implementation face different market pressures.
- Les Sabatini anticipates the expansion of the "virtual girlfriend economy" and digital doubles across various verticals, leading to numerous future moral quandaries.
- X estimates that the 2014 prediction regarding 80% of fourth-grade jobs becoming obsolete by age 18 is increasingly accurate due to AI advancement, while the push toward artificial general intelligence risks creating new markets that governments are unprepared to reskill for.
- Companies implementing AI without understanding boundaries face liability risks, specifically regarding lawsuits arising from chatbot hallucinations or the generation of false information.
- Chakrabarti suggests that traditional STEM education and coding instruction for children may become pointless within 10 years, predicting a curriculum shift as children naturally learn to write code.
- Near-term capabilities include AI reading DNA, a development described as frightening because humans currently lack the concept to verify the correctness of such outputs.
- Large language models are expected to shift toward generating smaller, controlled context boxes rather than processing entire context windows to manage error rates.
- Seth Ladelll predicts that for the next 12 to 24 months, AI will act as a "cape" enhancing human capabilities rather than replacing them entirely, necessitating a shift in education systems toward teaching the skill of learning new skills rather than lifelong static acquisition.
- Risks associated with AI mimicking reasoning without true comprehension include the potential for uncontrollable decisions being made before corrective action is possible.
- X forecasts that "super intelligence" will be reached when systems surpass human understanding of their own outputs, creating a moral quandary where control is lost.
- Department of Defense war gaming tests with ChatGPT are expected to continue revealing biased outputs, such as recommending nuclear strikes, reflecting a lack of true reasoning.
- Future regulation is anticipated to be necessary to protect humans within AI frameworks, potentially modeled after the EU's approach, though there is a risk that regulators remain unprepared due to a talent gap.
- Geopolitical AI development is forecasted to diverge, with the US and China aligning systems with respective value systems, with the US maintaining its constitutional rights while competing with China's model.
- The "AI policy lab" at UC Berkeley is expected to evolve into a living resource for policymakers to continuously update their knowledge as technology changes.
- Chakrabarti envisions a future where AI democratizes access and enables individuals to unlock human potential in fields like music and screenwriting without formal training, effectively evening the playing field for disadvantaged people.
- Sternberg remains optimistic that AI will democratize healthcare and education while serving as an enabler of global talent.
- Lau expects a future dialogue where the distinction between "AI native" and "AI wave rider" businesses blurs into a single conversation.
- Ladelll predicts that the definition of a good salesperson will shift from innate individual skill to the ability to leverage AI tools to elevate average performers to high levels.
- Ladelll anticipates that the current lag in teaching coding within education systems will result in children who can code being considered pointless in 10 years.