The AI Hype Cycle
Hyping the Hype for the Hype.
The Sentience Institute recently released an alarming report on the public's perception of AI in the United States. The study found that thirty-four percentĀ of the U.S. public believes that Artificial General Intelligence (AGI), a computer that can perform all the diverse tasks a human can do at a similar or higher level, already exists. Twenty-four percent of the respondents believe that an artificial "superintelligence," a machine that surpasses all human capabilities, already exists.
Neither AGI nor an Artificial superintelligence exists, and it is unclear if they ever will. All AI today is Narrow AI, a system optimized for a specific task that the computer can perform well if used only for the function it was designed and trained to do. All forms of AI today fit in this category, from large language models (LLMs) like ChatGPT to self-driving tractors and Netflix's recommendation system. A self-driving tractor might work great in a cornfield but not so great at recommending what to watch tonight.
In contrast, an AGI should be able to take a math test and then navigate a ship around the globe. Despite the hype you might have heard from tech CEOs and the fascinating results that ChatGPT confidently provides, AGI is entirely hypothetical. None exist. It will likely be decades before such systems exist, if they are even possible.
An Artificial Superintelligence is just science fiction, and there is no known pathway to move from narrow AI to a Superintelligence. Yet we see dire warnings about this type of AI taking over the world and treating us like pets at best and cockroaches at worst. Today, AI is more like a single-celled ameba than a super being.
Considering the hype around AI during the past five years, it is no wonder people think that AI is so much more advanced than it really is. OpenAI's CEO, Sam Altman, is one of the worst at hyping up AI's capabilities. He recently said AI could found and "make a great company" on its own, "know all of physics" and regularly talks as if AGI and computer superintelligence are just around the corner. For him, raising expectations raises the value of his company and personal wealth, which makes him more sought out for quotes about AI. It's a virtuous hype cycle of sorts that is creating a massive bubble of AI expectations and company valuations.
That valuation bubble is so great that Goldman Sachs recently warned about the vast investment big tech has made in AI infrastructure (data centers, power plants, and people) and the lack of any profitable applications for AI. The report "Gen AI: Too Much Spend, Too Little Benefit?" concluded that the industry would have to develop an application worth a trillion dollars in profit to just break even. And, of course, that number grows as time goes on.
Why would Tech leaders promote such hype? I suspect two reasons.Ā First, because of the Internet, monopolies dominateĀ the online platforms we use.Ā So, they are rushing to get the investment and interest needed to become "the next Google." If a company can develop an AGI, would the world need another? No, one would be plenty for the entire planet. And when the AI bubble pops, wiping out trillions of dollars in the stock market, each CEO wants their company to be the last company standing. Of course, this assumes there can be a winner, and the bubble doesn't just pop and take everyone down.
More nefariously, by hyping up the dangers of some possible future AI, CEOs can keep lawmakers and the public distracted from the severe damage that their Narrow AI systems are, or might, cause. For example, Narrow AI is theĀ type ofĀ algorithm that Instagram uses to keep sending messages about how to lose weight to girls with eating disorders (the goal of the Narrow AI, in this case, is to keep girls engaged on Instagram, not to promote a healthy lifestyle).
And CEOs visiting Capitol Hill would rather talk about a non-existent threat than why their companies lobby against data privacy bills that would give the public more say about how personal data is used, provide transparency into how their systems make decisions, recognize faces, and collect data, or discuss how their technology might find its way into the hands of our enemies.
So, don't worry about a Skynet system like in The Terminator or any other artificial "superintelligence." Instead, worry about why Narrow AI is getting stuffed into just about every application we use and who is stands to benefit.
From what I have seen so far, it is not us.
My commentary may be republished online or in print under Creative Commons license CC BY-NC-ND 4.0. I ask that you edit only for style or to shorten, provide proper attribution and link to my contact information.
š Upcoming Talks/Classes šØāš«
Workshop on Emerging Technologies for Digitalization
August 12, 2024 (TBC) | Lima, Peru
More information and will be available on the APEC Peru website.
Managing the Learning Machine
8:00 AM, September 10th. In Person and Zoom.
In this session, we will explore how AI, particularly ChatGPT and advanced machine learning technologies, is changing our world. We'll see how AI is making a big impact in different areas like medicine and retail, and how humans and machines can work together in new ways. AI can boost productivity but also come with risks like misinformation. This talk will help you understand the power and challenges of AI, and why ethical considerations are crucial as this technology continues to grow. Join us to learn how AI is reshaping industries and everyday life.
More information and Registration will be available on MU Retireeās Association website.
Technology Potpourri
9:30-11am, Sept 12, 19, 26, and Oct 3
This series will be taught by my friend and colleague, Sophia Rivera Hassemer, but I will get to be her assistant! It will be in person only at the Moss building, and will be very hands on with technology.
In this four part series, participants will gain hands-on experience emerging technology from the Center for Entrepreneurship and Innovation. Week one is GoGo Robot, where you learn how to program a robot car and explore the Internet of Things. During week two we will play with making 360 videos for youtube and Virtual Reality. Design objects for 3D printing and learn about rapid prototyping during week three. We will learn how to use AI to write computer code and excel functions during week four. Participants are encouraged to bring laptops or an iPad.
More information and Registration will be available on the Osher website.
ARTIFICIAL INTELLIGENCE AND THE ELECTIONS
Tuesday, September 10, 2024
6:30pm - 8:00pm
Artificial Intelligence (AI) is a technology that is changing and challenging the landscape of just about every field. J. Scott Christianson (aka Prof C) will explore AI's potential impact as a source of disinformation, especially during an election year. Along the way, he will provide some pointers for surviving and thriving in an AI-mediated world.
More information is available on the Missouri River Regional Library website.
Harnessing AI for Nonprofit Growth
10:45 - 11:45 a.m. November 7. On Zoom.
More information and Registration will be available on the New Chapter Coaching website.
AI: Current Trends and Future Directions
7:30 AM, November 12th. On Zoom.
Grab your coffee and get ready to review the significant progress made in generative AI, including its current applications and anticipated developments. Prof C will present recent research on how generative AI systems are utilized, focusing on PMI's initiatives and offerings. We'll discuss the implications of generative AI for project managers, highlighting practical use cases and best practices. The session will also touch on important InfoSec considerations.
Takeaways:
Understanding Progress: Gain insight into the advancements in generative AI and its current applications.
PMI Initiatives: Learn about PMI's initiatives and offerings in the generative AI space and how they can benefit your projects.
Future Developments: Anticipate future trends and developments in generative AI and how they might impact your work.
InfoSec Considerations: Understand important information security considerations related to generative AI to ensure safe and secure implementation.
Registration will be available on PMI Mid-MO Chapter's website.
Lovely and reassuring. Good points about the incentives of tech companies. And the tractor illustration--good fun!