Staring at the Blank Page Before You: Writing the Future of AI

Unit 1: Age of AIl; Topic 5: Staring at the Blank Page Before You: Writing the Future of AI

Jun 28, 2023

A quick note on how to engage with this article.

1. Like a class curriculum, each topic is part of a larger unit that dives into a distinct subject. Think of it like a vinyl record, each topic can be enjoyed as a track or fully experienced like an album. Subscribe to get them weekly in your inbox!

2. No knowledge gatekeeping here! We’re neophytes ourselves, presenting our laywomen’s research from topics that make us curious. If you see a bolded term, it is defined in the glossary at the end of this article.

3. Look out for underlined passages, these are your portal down the knowledge rabbit hole and lead to further reading. Insatiable? Original sources are merchandised on our interactive reference page at the end.

4. This is a reciprocal experience! We invite you to contemplate the discussion questions at the back. Or even cooler, forward the topic to a friend, colleague, or your crush and start a meaningful conversation.

Curiously,

Ambi & Abbey

Throughout our exploration of AI & the workplace, we gave a superficial overview of artificial intelligence, remembered how recent tech innovations shaped industry, and examined subsequent economic impacts on the worker/consumer. Through the rear window, we can muse about what the near future may hold, but it is ultimately unknowable.

Even ChatGPT concedes its own limitations. When prompted with questions about the future by hopeful fingertips, it monologues:

“As an AI language model, I can’t predict with certainty what will happen in the future.”

None of us are soothsayers, but it doesn’t take a Stanford dropout to predict that AI has the potential to scale both problems and solutions.

As with all things, there is a duplicity to AI’s power. Harnessed responsibly, it holds the potential to solve some of the most complicated global issues within the next decade. When wielded for the wrong motives, the decimating potential has been explored by nearly every prolific sci-fi writer.

To navigate this duality, society must be educated on both the risks and benefits, exploring the “dark side” of AI. To ask questions like: How can we use AI to produce the most good for the most people? How can we collectively plan for that future?

Sam Altman, who has been arguably one of the most involved leaders of the revolution, OpenAI’s CEO, has been attempting to lead conversations with global lawmakers around the regulation of AI, most recently with friction in Europe. But what do we as individuals need to be considering most?

The AI-Jump Scare The Robot in the Mirror

AI is data-hungry. Algos mature based on the data they consume: not innately “bad” or “evil”, outputs are mere extrapolated reflections. If algorithms are fed junk data, the input equivalent of Cosmic Brownies and Hot Cheetos, it’s going to be an unhealthy outcome.

While some may associate bias with opinion, bias shows up in a variety of clandestine ways. Bias is more than an additive slant, it is also skewed by the presentation of the data, sample size, data omissions, and the data’s “cleanliness”. To combat nuances of biased data, we need data hygienists of the future, as algo biases reinforced at scale can swiftly wreak havoc.

Through programming and natural language processing, all forms of AI take root from a human touchpoint.

No individual, or AI for that matter, is truly objective. Inputs from the human experience from our unique subjective reality, which develops around cognitive biases.

This is to say, AI is being trained by flawed humans with flawed data. OpenAI is written by a few minds and trained by the user base, through a process called Reinforcement Learning from Human Feedback (RLFHF).

Who among us humans is free from our own individual perspective? How do we neutralize the human factor? How can we trace artificial intelligence’s marionette strings up to accountable hands?

Beyond the Bias: The Outcome of the Inputs

While we’ve touched on the workplace effects of productivity technologies, innovations such as AI also have the potential to be both a poison and medicine with latent societal issues.

From misinformation to gaping holes in consumer protections, there are many lurking systemic cracks that if not carefully considered could have extrapolated, devastating quakes on society. Here are some such cracks, bound to be throttled by the siesmic quake of AI

Misinformation

Truth is a subject of ancient debate, its merits have been debated in the public forum from Plato to Trump.

Truth is a word that seems to become increasingly subjective in the era of 24/7 news cycles and unrelenting feeds of UGC, as our siloed online worlds are metaphysically real if not accurate.

Tech such as social media and the Internet has enabled the dissemination of unfounded “facts" on a global scale.

For all its abundance of peer-reviewed data and wealth of knowledge, the Internet is also teeming with cobwebs from outdated info and unfounded opinions from angry trolls. The web does not have a system for taking out the trash.

AI tools like ChatGPT learn autonomously by running on top of this patchwork of inputs, scraping the Internet far and wide to find trends amidst a plethora of topics, and misinformation’s threats compound.

As algos and humans alike can struggle to discern what’s garbage while surfing inputs from humans & the public domain, they can perpetuate fake news.

In code, this leads to what is known as “hallucinations,” or results that were not anticipated, do not match any discernible pattern, or are inconsistent with trained data.

The more trust we put into AI models to be objective truth-tellers, the more vulnerable we are to further distortions of reality when we act on their ill-founded conclusions.

Without proper education and regulation of AI interactions, we humans continue to struggle with discerning truth.

Social Surveillance & Privacy

Data-hungry tech like ChatGPT requires users – hella users – to offer queries, thoughts, and opinions, in order to reinforce learning.

As with social media, users are primarily concerned with the use-value for themselves (like handing off sales prospecting to an algo) than the data privacy (that your book or client info may be exposed).

The US has been lax behind the EU in terms of consumer data protection and now Congress is trying to wrap its head around regulation for both social media and AI.

Even with incoming freshmen, the Senate and Congress have a median age of 65.3 and 57.9 respectively. Voting seats are populated by non-digital natives who struggle to grasp the user perspectives of the tech they seek to regulate, as seen with the TikTok hearings. Lawmakers don’t need to understand the things they regulate, silly!

Manipulation & Data Poisoning

Because the integrity of AI’s outputs is maintained through the data it’s fed, a potential vulnerability is data poisoning and subsequent mind control of populations en masse.

As researchers from Cornell University explain, “Adversarial data poisoning is an effective attack against machine learning and threatens model integrity by introducing poisoned data into the training dataset.”

In a global setting, this poses many risks to shake up everything from political unrest to the stock market.

In May 2023, a dismal image of black smoke appearing to be arising from a government building near the Pentagon surfaced on social platforms, and moments after, the S&P saw a dip. It might have been the first time an AI-generated image moved the markets, according to Bloomberg.

If unregulated, these incidents may not be isolated. Market manipulation is a tangible instance of how data wars can shake out between nations, with information data a proxy to manipulate an entire population.

If trolls can make the internet believe Balenciaga-wearing Pope Francis was the collab of the century, imagine what an irate ex-politician or opposing nation is capable of.

Look on the Byte Side: AI for Good

As ChatGPT is helping millions of individuals outsource mundane tasks in their daily lives, the widespread implementation of ChatGPT and adjacent tech presents opportunities for collective advancement in key areas like business, the economy, and society as a whole.

On the Collective Level

Humanity is confronting some sticky situations. Climate change, an aging workforce, and medical care for the growing elderly population. AI can help lift the Herculean burden of these global challenges.

Climate Change

Climate change is one such hot topic (sorry, not funny) in the next decade. Scientists have been trying to wrap their arms around this issue for decades but have struggled to triangulate all the innumerable variables.

Because there are innumerable factors within a changing planet, we need additional support in accurately parsing through the data and finding nuanced trends.

For instance, researchers at George Washington University are leveraging machine learning to more accurately weigh the climate models used by the Intergovernmental Panel on Climate Change.

AI acts as a comrade in the battle for the planet, as it can monitor and categorize elements of climate change that are in constant motion, assisting humans to make better predictions about alterations in the environment and therefore take informed action sooner.

Birth Rates

Falling birth rates and aging populations pose another impending challenge. In many economies, labor productivity, a primary influencer for economic growth, has slowed; it dropped an average of 0.5% in 2010-2014 from 2.5% in the previous decade in the U.S. and major European economies.

Tech, such as artificial intelligence, can bolster the productivity index in the wake of a waning workforce. AI and automation do have the potential to help us reach 2% annually within the next decade.

Deploying artificial neural networks could be responsible for 3.5-3.8 trillion in annual revenue, or 40% of the value created by all analytics techniques. While this does not solve the dilemma of falling birth rates, it certainly could help to lighten the stork’s load.

Medical Research

Medical research has promising applications for AI, an area historically throttled by accessibility, misdiagnosis, and equity.

From more accurate cancer diagnoses helping treat patients sooner to virtual health assistants helping patients in remote or conflict-ridden locations, AI in the healthcare sector is one that shows early promise.

As the National Institutes of Health explains, while computer systems often execute tasks more efficiently than humans, more recently, state-of-the-art computer algorithms have achieved accuracies that are at par with human experts in the field of medical sciences.

Food Insecurity

Food insecurity & the global marketplace is intimately related to climate change. Billions of people are overweight while billions starve, a misallocation of resources.

30-40% of all food produced is lost in the supply chain and the energy put into the global distribution of food equates to 3.3 billion metric tonnes of carbon dioxide. What if we were to use artificial intelligence to help distribute our existing resources with the least amount of pollution?

On the Individual & Sheerly Human Level

Let’s recall the skills that differentiate the woman from the machine.

Social skills, such as providing emotional support to someone grieving
Unpredictable physical skills, such as elderly care or replacing a tire on the side of a road
Common sense, such as knowing that someone walking quickly is likely in a rush
General intelligence, such as hearing yelling and knowing to call 911

AI is currently not as equipped at doing what we do best – thinking, feeling, inferring, musing, and reacting. AIgos shine with repetitive tasks, finding trends in masses of data, and drawing conclusions. Tasks often rob us of much time but also mundane mental space. What we do with the outputs of AI is another consideration altogether.

Artificial intelligence will extrapolate our human flaws, but if navigated with compassion, caution, and intention, there is a world where AI unlocks more of our human potential.

In the context of a career, automation tools already offload a lot of BS. From scheduling emails to high-level research to setting reminders, AI is an expert admin. The details that sneakily siphon away our focus in a single day and clog our inbox could be unburdened.

Lightening our workload, AI can carve out space for creativity, leadership, and deeper emotional fulfillment.

Beyond labor productivity, AI could spawn more opportunities to slow down and connect with the human minds around us, creating a positive reinforcement loop of innovation and previously imaginable speed.

In order to self-actualize, we must have our basic needs such as psychological (food, shelter) and community to be met to be our most creative.

This phenomenon is known as Maslow’s Hierarchy of Needs. If AI can help us maximize resources, like food wastage, to meet the funadmental needs of a global population, more individuals can realize their potential.

Every technology era has been followed by a proclamation of an Age of Leisure, one that has never inevitability materialized. In 1930, in the wake of the industrial revolution, British economist John Maynard Keynes famously predicted that by 2030, his grandchildren would be working 15 hours per week.

Keynes’ fantasy that money would be less of a consideration for his lineage has yet to come true - the average American worked just over 8 hours per day, or 40 per week, in 2021.

This is because technology, rather than maintaining productivity and alleviating labor, tends to spawn a period of economic growth.

The nation’s productivity index gets pumped by the new, productive technologies and the growth spawns new needs, new markets, and new opportunities for commercialization.

Lifestyle creep feeds consumer appetites in the West, putting us in a cycle of consumption and labor. Without intervention, ChatGPT and similar technologies will rationally perpetuate these trends.

Superminds

It’s not a matter of people or machines when we look at the future of work, but rather people and machines, as the two may eventually converge.

Integrated AI and humans, coined superminds, could be on the horizon, a sci-fi tale of bionic, superhuman ability that stems from the harmony of AI and human ability.

By combining the strengths of AI and humanity to complement one another, superminds promise a symbiosis that would unlock untold potential and efficiency.

The supermind optimists paints a halcyonian landscape, one in which technology and humans are in harmony to reach new heights, but if this integrated supermind tech were to be accessible strictly to those with wealth, it would vehemently exacerbate inequalities. Such advantages must be considered through an ethical lens, as with eugenics.

The Future is Being Written

The differentiating factor between AI and previous technological eras is sheer speed. Over the course of crafting this article, major shifts in the industry as a result of AI materialized.

As we crafted this article you are reading currently, there was an unrelenting stream of headlines detailing developments in AI tech.

Articles like “Programmer hooks up ChatGPT to a Furby and the toy makes a terrifying claim it will ‘infiltrate households’” to “Discord chatbot shares napalm instructions” to “Musk wants to develop TruthGPT,” rapid-fire at rates faster than Elon’s Twitter fingers.

Humans tend to fear the unknown and it can often be confusing where to look for answers while treading uncertain waters.

Tech anxiety must not lend itself to despair, a state of inaction where we forfeit our collective agency to shape to future.

Within the sea of fatalist headlines and rose-colored claims, one thing's for sure…the future is still being written. And you are holding the pen.

Let’s exhale our fears and start tugging at productive questions such as, How does this technology fit into my life? How can I prepare for a world where automation does parts of my job better than me? What parts of my life do I wish I didn’t have to do?

And outside of ourselves…How can leaders better acknowledge the pace at which this technology is evolving? Who is valuing privacy and integrity along the way? How are we creating safe holds against our bias being automated and deeply ingrained?

In the end, we are a deeply beautiful, flawed, and creative species. Deux Ex Machina is only a reflection of the human playing God. AI holds up the mirror, what do we want it to reflect?

Discussion Questions:

What do you think the impact and legacy of ChatGPT will be from a global standpoint?
What do you think is the most pressing issue artificial intelligence can help us solve?
How do you think lawmakers should approach the regulation of artificial intelligence technologies?
The implications of AI are nuanced & faceted, we only touched on a select few angels in our weeks-long exploration. What other potential implications of AI can you think of?

Glossary:

Cognitive Bias: a systematic error in thinking that occurs when people are processing and interpreting information in the world around them and affects the decisions and judgments that they make.

Data Poisoning: Refers to injecting malicious or misleading data into a dataset for training machine learning models. This compromises model performance and integrity, resulting in inaccurate or biased predictions. The goal is to manipulate the learning process and exploit vulnerabilities in the training algorithm.

Maslow’s Hierarchy of Needs: Maslow's Hierarchy of Needs is a psychological theory proposed by Abraham Maslow in the field of human motivation and behavior. It suggests that human needs can be organized into a hierarchical structure, with basic physiological needs at the bottom and higher-level needs at the top.

Subjective Reality: Subjective reality refers to an individual's unique and personal experience of reality, shaped by their perceptions, thoughts, emotions, and beliefs. In subjective reality, individuals construct their own understanding of the world based on their internal cognitive and emotional processes.

Reinforcement Learning from Human Feedback (RLFHF): Reinforcement Learning from Human Feedback (RLFHF) is an approach in the field of artificial intelligence and machine learning that combines reinforcement learning techniques with human guidance and feedback to train intelligent systems. RLFHF aims to improve the learning process by incorporating human expertise and knowledge into the training loop.

UGC (User Generated Content): User-generated content (UGC) refers to any form of content, such as text, images, videos, reviews, comments, or other media, that is created and shared by users or consumers of a product, service, or platform.