mitsloan.mit.edu Open in urlscan Pro
2620:12a:8000::3  Public Scan

Form analysis
2 forms found in the DOM

<form id="nav-search-form" class="nav--search-form" data-once="form-updated" data-drupal-form-fields="">
  <input tabindex="-1" class="nav--search-field" aria-label="placeholder" type="text" placeholder="Search">
  <input tabindex="-1" type="submit" class="js-nav-search-submit nav--search-submit" value="Submit">
  <span class="nav--search-submit-graphic" aria-hidden="true"></span>
</form>

POST /ideas-made-to-matter/why-neural-net-pioneer-geoffrey-hinton-sounding-alarm-ai

<form
  class="webform-submission-form webform-submission-add-form webform-submission-emma-subsc-form webform-submission-emma-subsc-add-form webform-submission-emma-subsc-paragraph-19375-form webform-submission-emma-subsc-paragraph-19375-add-form js-webform-details-toggle webform-details-toggle"
  data-drupal-selector="webform-submission-emma-subsc-paragraph-19375-add-form-2" action="/ideas-made-to-matter/why-neural-net-pioneer-geoffrey-hinton-sounding-alarm-ai" method="post" id="webform-submission-emma-subsc-paragraph-19375-add-form--2"
  accept-charset="UTF-8" data-once="form-updated" data-drupal-form-fields="edit-email--2,edit-actions-submit--2,edit-secondaryemail--2">
  <label data-drupal-selector="edit-work-smart-with-our-thinking-forward-newsletter" for="edit-work-smart-with-our-thinking-forward-newsletter--2">Work smart with our Thinking Forward newsletter</label><label
    data-drupal-selector="edit-insights-from-mit-experts-delivered-every-tuesday-morning-" for="edit-insights-from-mit-experts-delivered-every-tuesday-morning---2">Insights from MIT experts, delivered every Tuesday morning.</label>
  <div class="js-form-item form-item js-form-type-email form-item-email js-form-item-email">
    <input autocomplete="email" pattern="^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$" data-webform-pattern-error="Not a valid email" data-drupal-selector="edit-email" type="email" id="edit-email--2" name="email" value="" size="60" maxlength="254"
      placeholder="*" class="form-email required" required="required" aria-required="true"><label for="edit-email--2" class="option js-form-required form-required">Email Address</label>
  </div>
  <div class="macro-cta__btn js-form-wrapper form-wrapper" data-drupal-selector="edit-submit-buttons-container" id="edit-submit-buttons-container--2">
    <div data-drupal-selector="edit-actions" class="form-actions webform-actions js-form-wrapper form-wrapper" id="edit-actions--3"><input class="webform-button--submit button button--primary js-form-submit form-submit"
        data-drupal-selector="edit-actions-submit-2" type="submit" id="edit-actions-submit--2" name="op" value="SIGN UP"></div>
    <div class="arrow js-form-item form-item js-form-type-webform-markup form-item- js-form-item- form-no-label">
    </div>
  </div>
  <div class="webform-footer js-form-wrapper form-wrapper" data-drupal-selector="edit-footer" id="edit-footer--4">
    <div class="footer-image thinking-forward--plus-signs js-form-item form-item js-form-type-webform-markup form-item- js-form-item- form-no-label">
    </div>
  </div>
  <input data-drupal-selector="edit-honeypot-time" type="hidden" name="honeypot_time" value="SuHkDSf6MfekPRclFeNJFFFT--mBp5t9e-pNT8lE4VQ"><input autocomplete="off" data-drupal-selector="form-hqet63qvze18tszo8vehpz6sx4j7dldq89qd7ai5mks" type="hidden"
    name="form_build_id" value="form-hqEt63qVzE18tszo8vehpz6SX4J7dLdq89qD7Ai5mks"><input data-drupal-selector="edit-webform-submission-emma-subsc-paragraph-19375-add-form-2" type="hidden" name="form_id"
    value="webform_submission_emma_subsc_paragraph_19375_add_form">
  <div class="secondaryemail-textfield js-form-wrapper form-wrapper" style="display: none !important;">
    <div class="js-form-item form-item js-form-type-textfield form-item-secondaryemail js-form-item-secondaryemail">
      <label for="edit-secondaryemail--2">Leave this field blank</label>
      <input autocomplete="off" data-drupal-selector="edit-secondaryemail" type="text" id="edit-secondaryemail--2" name="secondaryemail" value="" size="20" maxlength="128" class="form-text">
    </div>
  </div>
</form>

Text Content

Skip to main content
MIT Sloan logo
 * academics
 * Ideas Made to Matter
 * Diversity
 * Events
 * Alumni
 * Faculty
 * About
 * Executive Education
 * academics
 * contact

 * facebook
 * instagram
 * linkedin
 * twitter
 * youtube

contact
Search Show Main Navigation

Smart. Open. Grounded. Inventive. Read our Ideas Made to Matter.


close search
Back to Menu


WHICH PROGRAM IS RIGHT FOR YOU?

MBA

Through intellectual rigor and experiential learning, this full-time, two-year
MBA program develops leaders who make a difference in the world.

Master of Finance

A rigorous, hands-on program that prepares adaptive problem solvers for premier
finance careers.

Master of Business Analytics

A 12-month program focused on applying the tools of modern data science,
optimization and machine learning to solve real-world business problems.

Leaders for Global Operations

Earn your MBA and SM in engineering with this transformative two-year program.

Master of Science in Management Studies

Combine an international MBA with a deep dive into management science. A special
opportunity for partner and affiliate schools only.

PhD

A doctoral program that produces outstanding scholars who are leading in their
fields of research.

Undergraduate

Bring a business perspective to your technical and quantitative expertise with a
bachelor’s degree in management, business analytics, or finance.

System Design & Management

A joint program for mid-career professionals that integrates engineering and
systems thinking. Earn your master’s degree in engineering and management.

Integrated Design & Management

An interdisciplinary program that combines engineering, management, and design,
leading to a master’s degree in engineering and management.


EXECUTIVE PROGRAMS

MIT Sloan Fellows MBA

A full-time MBA program for mid-career leaders eager to dedicate one year of
discovery for a lifetime of impact.

MIT Executive MBA

This 20-month MBA program equips experienced executives to enhance their impact
on their organizations and the world.

Executive Education

Non-degree programs for senior executives and high-potential managers.

Visiting Fellows

A non-degree, customizable program for mid-career professionals.

View Upcoming Program Events


RECENT

14 hours ago

Johnson & Johnson’s 3-step strategy for innovation

Jul 24, 2023

Northwestern Mutual’s strategy chief: Take on challenges others won't

Jul 20, 2023

How to develop a carbon management strategy for business

Credit: Mimi Phan / Cole Burston via Getty Images



Ideas Made to Matter

Artificial Intelligence


WHY NEURAL NET PIONEER GEOFFREY HINTON IS SOUNDING THE ALARM ON AI

By

Sara Brown

May 23, 2023


WHY IT MATTERS

Geoffrey Hinton, a respected researcher who recently stepped down from Google,
said it’s time to confront the existential dangers of artificial intelligence.

Share
 * 
 * 
 * 
 * 
 * 

 * facebook
 * twitter
 * linkedin
 * email
 * print

open share links close share links

A deep learning pioneer is raising concerns about rapid advancements in
artificial intelligence and how they will affect humans.

Geoffrey Hinton, 75, a professor emeritus at the University of Toronto and until
recently a vice president and engineering fellow at Google, announced in early
May that he was leaving the company — in part because of his age, he said, but
also because he’s changed his mind about the relationship between humans and
digital intelligence.

Work smart with our Thinking Forward newsletterInsights from MIT experts,
delivered every Tuesday morning.
Email Address


Leave this field blank

In a widely discussed interview with The New York Times, Hinton said generative
intelligence could spread misinformation and, eventually, threaten humanity. 

Speaking two days after that article was published, Hinton reiterated his
concerns. “I’m sounding the alarm, saying we have to worry about this,” he said
at the EmTech Digital conference, hosted by MIT Technology Review.

Hinton said he is worried about the increasingly powerful machines’ ability to
outperform humans in ways that are not in the best interest of humanity, and the
likely inability to limit AI development.


THE GROWING POWER OF AI

In 2018, Hinton shared a Turing Award for work related to neural networks. He
has been called “a godfather of AI,” in part for his fundamental research about
using back-propagation to help machines learn.  

> I think it’s quite conceivable that humanity is just a passing phase in the
> evolution of intelligence.

Geoffrey Hinton Former Vice President and Engineering Fellow, Google
Share
 * 
 * 
 * 
 * 
 * 

Hinton said he long thought that computer models weren’t as powerful as the
human brain. Now, he sees artificial intelligence as a relatively imminent
“existential threat.”

Computer models are outperforming humans, including doing things humans can’t
do. Large language models like GPT-4 use neural networks with connections like
those in the human brain and are starting to do commonsense reasoning, Hinton
said.

These AI models have far fewer neural connections than humans do, but they
manage to know a thousand times as much as a human, Hinton said.

In addition, models are able to continue learning and easily share knowledge.
Many copies of the same AI model can run on different hardware but do exactly
the same thing.

“Whenever one [model] learns anything, all the others know it,” Hinton said.
“People can’t do that. If I learn a whole lot of stuff about quantum mechanics
and I want you to know all that stuff about quantum mechanics, it’s a long,
painful process of getting you to understand it.”

AI is also powerful because it can process vast quantities of data — much more
than a single person can. And AI models can detect trends in data that aren’t
otherwise visible to a person — just like a doctor who had seen 100 million
patients would notice more trends and have more insights than a doctor who had
seen only a thousand.  


AI CONCERNS: MANIPULATING HUMANS, OR EVEN REPLACING THEM

Hinton’s concern with this burgeoning power centers around the alignment problem
— how to ensure that AI is doing what humans want it to do. “What we want is
some way of making sure that even if they’re smarter than us, they’re going to
do things that are beneficial for us,” Hinton said. “But we need to try and do
that in a world where there [are] bad actors who want to build robot soldiers
that kill people. And it seems very hard to me.”

Humans have inherent motivations, such as finding food and shelter and staying
alive, but AI doesn’t. “My big worry is, sooner or later someone will wire into
them the ability to create their own subgoals,” Hinton said. (Some versions of
the technology, like ChatGPT, already have the ability to do that, he noted.)

“I think it’ll very quickly realize that getting more control is a very good
subgoal because it helps you achieve other goals,” Hinton said. “And if these
things get carried away with getting more control, we’re in trouble.”

Artificial intelligence can also learn bad things — like how to manipulate
people “by reading all the novels that ever were and everything Machiavelli ever
wrote,” for example. “And if [AI models] are much smarter than us, they’ll be
very good at manipulating us. You won’t realize what’s going on,” Hinton said.
“So even if they can’t directly pull levers, they can certainly get us to pull
levers. It turns out if you can manipulate people, you can invade a building in
Washington without ever going there yourself.”

At worst, “it’s quite conceivable that humanity is just a passing phase in the
evolution of intelligence,” Hinton said. Biological intelligence evolved to
create digital intelligence, which can absorb everything humans have created and
start getting direct experience of the world. 

“It may keep us around for a while to keep the power stations running, but after
that, maybe not,“ he added. “We’ve figured out how to build beings that are
immortal. These digital intelligences, when a piece of hardware dies, they don’t
die. If … you can find another piece of hardware that can run the same
instructions, you can bring it to life again. So we’ve got immortality, but it’s
not for us.”


BARRIERS TO STOPPING AI ADVANCEMENT

Hinton said he does not see any clear or straightforward solutions. “I wish I
had a nice, simple solution I could push, but I don’t,” he said. “But I think
it’s very important that people get together and think hard about it and see
whether there is a solution.”

More than 27,000 people, including several tech executives and researchers, have
signed an open letter calling for a pause on training the most powerful AI
systems for at least six months because of “profound risks to society and
humanity,” and several leaders from the Association for the Advancement of
Artificial Intelligence signed a letter calling for collaboration to address the
promise and risks of AI.


RELATED ARTICLES

It’s not too late to rechart the course of technology The AI road not taken MIT
Sloan research on AI and machine learning

It might be rational to stop developing artificial intelligence, but that’s
naive and unlikely, Hinton said, in part because of competition between
companies and countries.

“If you’re going to live in a capitalist system, you can’t stop Google [from]
competing with Microsoft,” he said, noting that he doesn’t think Google, his
former employer, has done anything wrong in developing AI programs. “It’s just
inevitable in the capitalist system or a system with competition between
countries like the U.S. and China that this stuff will be developed,” he said.

It is also hard to stop developing AI because there are benefits in fields like
medicine, he noted.  

Researchers are looking at guardrails for these systems, but there is the chance
that AI can learn to write and execute programs itself. “Smart things can
outsmart us,” Hinton said.

One note of hope: Everyone faces the same risk. “If we allow it to take over, it
will be bad for all of us,” Hinton said. “We’re all in the same boat with
respect to the existential threat. So we all ought to be able to cooperate on
trying to stop it.”

Study: Industry now dominates AI research 

For more info Sara Brown Senior News Editor and Writer sbrown1@mit.edu


RELATED ARTICLES

Ideas Made to Matter Johnson & Johnson’s 3-step strategy for innovation Ideas
Made to Matter How to set technology strategy in the age of AI Ideas Made to
Matter The world needs a global AI observatory — here’s why


THE MISSION

The mission of the MIT Sloan School of Management is to develop principled,
innovative leaders who improve the world and to generate ideas that advance
management practice.


FIND US

MIT Sloan School of Management 100 Main Street Cambridge, MA 02142 617-253-1000


LINKS

 * Keeping MIT Healthy
 * Press
 * Careers
 * Accessibility
 * Licensing
 * Privacy


©2023 MIT Sloan School of Management




previousnextstart slideshow