Afternoon Audit

Share this post
A Seriously Realistic Technological Threat to Society
www.afternoonaudit.co

A Seriously Realistic Technological Threat to Society

Thursday, March 11th (60 Second Read)

Afternoon Audit
Mar 11, 2021
Comment
Share

Equities Rally Higher as Stimulus Bill Becomes Law

Dow: +0.58% | Nasdaq: +2.52% | S&P: +1.04%

Catch Up Quick

  • President Biden has signed the $1.9T coronavirus relief bill, which notably includes $60B of tax hikes for the wealthy + large corporations

  • The total economic stimulus passed by the U.S. government over the past year has reached ~$5.3T → that’s >$43K per U.S. household (Axios Capital)

  • Economists revise growth estimates of the U.S. economy to a 6% pace this year (up from 4.3%), forecasting an addition of ~3M new jobs (WSJ)

  • A Deutsche Bank survey revealed $150B in stimulus could go into stocks, as retail investors plan to invest 37% of payments

  • The Dow has surged to record highs, closing well above 32K

  • Social videogame platform Roblox launched its IPO yesterday at a valuation of roughly $40B

  • The U.S. is averaging fewer than 50K new coronavirus cases per day for the first time since October 2020

  • Coupang, a South Korean eCommerce giant, raised $4.6B in its IPO, implying an initial market cap of ~$60B

  • A SurveyMonkey poll revealed only 4% of ~4K respondents said that they've heard / read "a lot" about Clubhouse → TikTok equivalent was 40%

  • According to new research, energy-intensive indoor marijuana growth has become a sizable source of greenhouse gas emissions (Axios)

  • Global food prices rose by a staggering 2.4% in February (The Food and Agriculture Organization's Food Price Index)

  • The Senate confirmed Merrick Garland as Attorney General


Thought of the Day

  • Over the past few weeks, “deepfake” videos of actor Tom Cruise circulated virally across TikTok and other platforms, raising both awe and concern alike

    The term deepfake refers to a video, photo, or audio recording that appears real but has been artificially constructed / manipulated to replace faces, manipulate expressions, and synthesize speech

  • Though appearing highly realistic, Tom Cruise was not involved in any way, shape, or form in the creation of the eerie video below:

  • As you can imagine, the danger of this increasingly convincing technology creates a plethora of opportunity for malicious intent → the direct threat to world leaders / celebrities + the eroding trust in crucial video evidence in legal processes barely scratch the surface

  • While we have previously covered A.I (such as here and here), the deepfake angle arguably poses its largest current threat to society at this point in time

  • To grasp the technology at work here → neural networks, the backbone of machine learning, are essentially computational systems made up of many algorithmic components that collectively process information similar to how the human brain does

  • Staying out of the weeds, the basic idea with machine learning modules that take images as input, is that each layer (or algorithm / node) in the neural network will decipher increasingly complex features

  • For instance, a highly simplified abstraction of an image recognizing neural network → 1) pixel pooling and convolution algorithms compress an image from a data storage perspective without removing key features (and in some cases, enhancing key features) 2) the next algorithm in the network might detect edges of the image (i.e. where an eye meets an eyelid) 3) the next algorithm might detect specific facial features (i.e. large blue eyes) 4) the next algorithm detects entire images (i.e. aggregating previous discoveries in a hierarchal fashion to recognize an image, using a feedback loop if incorrect to continuously learn & improve accuracy)

  • Deepfakes are an exponentially complex application of machine learning-based image recognition, in which modules are extensively trained with picture, video, and audio data to the point where inputs (any actions and words one could think of) can be turned into outputs (video and audio) using various algorithmic learning techniques that leverage the data the system has been trained with

The Bottom Line

  • Ultimately, as the software improves, it is critical that parallel technologies are developed and widely utilized to identify and expose illegitimate videos, which in turn infuses the security software realm with copious opportunities of huge value proposition!

Home

CommentComment
ShareShare

Create your profile

0 subscriptions will be displayed on your profile (edit)

Skip for now

Only paid subscribers can comment on this post

Already a paid subscriber? Sign in

Check your email

For your security, we need to re-authenticate you.

Click the link we sent to , or click here to sign in.

TopNewCommunity

No posts

Ready for more?

© 2022 Afternoon Audit
Privacy ∙ Terms ∙ Collection notice
Publish on Substack Get the app
Substack is the home for great writing