Are AI’s promises of high productivity gains realistic, given the current technological limitations?
AI, with its generative capabilities, promises to automate tasks, enhance productivity, and spur economic growth. However, the road to achieving these lofty goals is fraught with obstacles.
According to Goldman Sachs, the tech industry’s projected $1 trillion investment in AI over the next few years has yet to yield large returns beyond modest efficiency gains. Even NVIDIA, a major player in the AI hardware sector, has seen its stock price corrected sharply.
The skepticism surrounding AI’s potential is echoed by experts. Take Daron Acemoglu from MIT, for instance. He’s quite skeptical about AI’s grand promises.
According to him, only a small fraction of work tasks will be automated by AI in the next decade, impacting less than 5% of all tasks.
He argues that AI won’t advance fast enough to make a large economic impact, predicting only a slight bump in productivity and GDP growth. It’s a far cry from the revolution many envision.
On the other side of the spectrum, Joseph Briggs from Goldman Sachs is more optimistic. He believes that eventually, AI will automate 25% of tasks, leading to high productivity gains.
His optimism is based on the idea that costs will eventually decline, making AI automation more affordable. However, this pov is tempered by the reality of high initial costs and technical hurdles that need to be overcome.
Combining AI and crypto brings a new set of challenges. AI’s need for massive computational power and the current limitations in chip supply could slow down progress for crypto-based AI applications. Cryptographic algorithms used in AI-driven crypto solutions require extensive processing power, which could be hampered by these limitations.
Additionally, the power demands for AI data centers are skyrocketing, putting a strain on our already aging power grids. This is particularly critical for crypto mining operations, which are already under scrutiny for their high energy consumption.
Experts like Brian Janous, former VP of Energy at Microsoft, warn that our infrastructure isn’t ready for this surge in demand, which could stifle the growth of both AI and AI-enhanced crypto technologies.
So, where does this leave us? Are AI and crypto on the brink of bursting their bubbles, or are we witnessing the early stages of a technological breakthrough? Let’s find out.
AI-crypto bubble in making?
In recent discussions, leading experts Daron Acemoglu from MIT and Jim Covello, Head of Global Equity Research at Goldman Sachs, have shared insights on the future of AI and its economic impacts.
Acemoglu is cautious about AI’s economic potential, predicting a modest 0.5% increase in productivity and a 1% increase in GDP over the next decade, which contrasts starkly with Goldman Sachs’ more optimistic forecast of a 9% productivity boost and a 6.1% GDP increase.
Acemoglu’s skepticism stems from the current focus on generative AI, which primarily automates specific tasks rather than transforming entire industries.
For instance, AI might improve efficiency in tasks like data analysis but won’t largely impact multifaceted tasks requiring real-world interaction, such as those in transportation and manufacturing.
He estimates that AI could cost-effectively automate only 4.6% of all tasks in the next decade, translating to a 0.66% increase in total factor productivity and a 0.9% boost in GDP.
Covello shares Acemoglu’s skepticism but from a different angle. He highlights the substantial costs associated with AI infrastructure, which are projected to exceed $1 trillion in the coming years. Covello questions whether AI can solve complex and important problems at a cost that justifies such an investment.
Drawing parallels with past technology transitions, he notes that AI’s high costs and uncertain ROI set it apart from previous innovations like the internet, which provided low-cost solutions from the start.
Covello also challenges the assumption that AI costs will decline speedily over time. He points out that the AI hardware market is currently dominated by Nvidia, with little competition to drive down prices.
Nvidia’s monopoly, coupled with the immense initial costs of AI infrastructure, raises doubts about the technology becoming affordable enough for widespread adoption.
Despite these reservations, the AI and crypto intersection is an area of immense interest and potential.
Throughout 2024, AI-related crypto tokens have surged in popularity, with their combined market cap reaching around $30 billion as of July 23. This hype is driven by the belief that AI can innovate various sectors, including finance and healthcare.
Moreover, three prominent AI and crypto companies — Fetch.ai, SingularityNET, and Ocean Protocol—have recently formed an alliance, merging their tokens to create a new AI token called Artificial Superintelligence (ASI).
Their vision is to develop a decentralized AI platform aiming for Artificial General Intelligence (AGI) and eventually Artificial Superintelligence (ASI).
AGI refers to AI systems capable of performing any human task with the same level of competence, while ASI surpasses human capabilities entirely.
However, the journey to ASI is fraught with challenges. The need for massive computational power and the limitations in chip supply could slow down progress for AI-driven crypto applications.
Additionally, the strain on power grids due to the high energy demands of AI data centers and crypto mining operations cannot be overlooked.
You might also like: 5 AI coins to consider as SingularityNet and Worldcoin surge
A union in progress
Recent developments explain how closely aligned AI and crypto are and how they could become more aligned in the coming years.
For instance, Bitcoin miners, facing reduced rewards due to the Bitcoin halving event, are moving towards AI. The recent halving event in April 2024 reduced mining rewards from 6.25 Bitcoin to 3.125, pushing companies like Lancium and Crusoe Energy Systems to invest in AI data centers.
Their multibillion-dollar deal to build a 200-megawatt data center in Texas aims to meet AI’s growing demands, signaling a shift away from traditional Bitcoin mining.
Why is this happening? Bitcoin mining and AI infrastructure share common ground. Both require extensive data centers, high energy consumption, and advanced cooling systems.
As AI’s computational needs grow, Bitcoin miners see an opportunity to repurpose their existing infrastructure. Companies like Core Scientific and Hut 8 have already expanded into AI, betting on its long-term potential.
Adding another layer to this intersection is Grayscale’s recent launch of a digital asset fund focused on AI tokens, indicating skyrocketing demand.
We’re pleased to announce the launch of our newest private placement investment product, Grayscale Decentralized AI Fund. Available to eligible accredited investors.$NEAR $RNDR $FIL $LPT $TAOSee important disclosures or speak directly to a team member: https://t.co/gYetdms280 pic.twitter.com/OvNhec95io
— Grayscale (@Grayscale) July 17, 2024
So, where does this leave us? Are AI and crypto on the verge of a breakthrough or collaboration, or are we merely inflating a bubble?
The answer lies in balancing optimism with realism. Both technologies hold immense potential, but their success hinges on overcoming technical, financial, and infrastructural challenges.
Given the era we are living in, the possibilities are endless. On one end of the spectrum, we could see potential fusion and growth in both industries, or perhaps a complete overhaul should the circumstances change.
Whatever the case, the tech era has taken the world by storm, and unexpected changes are certainly on the horizon in the coming years and decades.
You might also like: Open letter warns about AI — and it should apply to crypto too