Things Are Getting Real
As AI gets more powerful, it’s starting to connect with our feelings, too.
Six months have passed since The Future of Life Institute published an open letter specifically noting that this new wave of AI systems is becoming “human-competitive at general tasks” and therefore, we should be asking ourselves such questions as: “Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization?”
To give us all a chance to ask and answer these questions, the letter called for a pause for all developments of this technology until we can set up new institutions and infrastructure to manage them.
And? Suffice it to say, nothing has been paused. The letter had some big signatories such as Elon Musk, Steve Wozniak and Andrew Yang (however, it should be noted that verification of some of the other names is in dispute). But when asked, some of the signatories basically admitted that they never expected it to work – and some even explicitly continued doing the type of work that they asked to be paused. During this time, technology continued to not only move quickly: it accelerated.
Case in point? ChatGPT can now see, hear and speak. It can also browse the web (again).
We are all familiar with chatbots at this point. But one that can see? And hear? It's actually pretty mind-blowing. You now never have to decode the craziness of parking signs again: just point your camera, upload the photo to ChatGPT and ask it, “Can I park at this spot right now?” Or show it a picture of a human cell and ask it to explain it to you like an eighth grader. For extra excitement, turn on “voice conversations” and literally just talk to ChatGPT and ask it these questions. If you squint real hard, you could imagine you’re just talking to a friend on the other end of a phone call.
Web browsing significantly enhances ChatGPT. I asked, "Did congress manage to fund the government?" and watched it as it consulted Bing, then Reuters, before informing me about the temporary 45-day funding measure. When I inquired about the implications for Speaker McCarthy, it said Kevin McCarthy has been ousted as Speaker of the House following a vote. This is all news as it happens, and a far cry from when it used to complain about its training data ending in September 2021.
Take my word for it, these are all foundational shifts and it's really easy to future-cast around it. Alan Turing proposed a game in 1950 as a way of evaluating whether a machine exhibits intelligent behavior that is equivalent, or indistinguishable, from that of a human. There is some debate whether ChatGPT can pass the Turing test, or whether we need to update that test, but, no doubt, these systems can now seem very human-like. Look, humans have the ability to anthropomorphize a rock – and now that “rock” has the power of language. People are going to start building emotional connections to these machines.
We are still in the era of move fast and break things. But what happens when those things that might be broken are our emotions?
Last week, I mentioned a post from one of OpenAI’s safety engineers comparing talking to ChatGPT as therapy. I can’t shake that post from my mind. It reminded me of a great quote from Episode Six of the podcast,
when I spoke with Roz Picard of the MIT Media Lab, the creator of affective computing. She told me that chatbots “are trained to respond to conversational elements, but that’s not the same as being a therapist: Does it know what it's saying? No. Does it know that it's a good thing? It only knows it got a positive reward mechanism…but it doesn't know your feelings. It doesn't care about you. It doesn't know what matters to you. It is really a smart parrot that has been taught a collection of things that it's been told to say in response to what you say.”
AI has many applications in managing emotions and mental health, like addressing call center burnout. Through affective computing, it's possible to gauge callers' emotional states, ensuring that distressed callers are spread evenly among agents rather than overwhelming a few. This balance allows agents to have emotional capacity to feel empathy for each caller that calls in. However, this raises concerns about AI's role in handling our emotions. What safeguards or values should be established to navigate this?
The initial launch of ChatGPT revealed a lack of safeguards, and it was quickly disabled due to enabling paywall circumvention, irking publishers. But that’s simply an economic side effect. This time, it's our headspace at risk. And this isn’t a theoretical risk – we know it's possible. In 2012, Meta (then Facebook) conducted an experiment to manipulate user emotions via NewsFeed adjustments.
Dr. Aboujaoude, a Stanford University professor and psychiatrist, discusses the trade-off between social media engagement and deeper personal relationships. Now, the advent of AI tools like chatbots, designed with “positive reward mechanisms,” tempts us with convenient interactions, possibly at the cost of genuine human connection (reminds me of the movie “Her”, “You always wanted to have a wife without the challenges of actually dealing with anything real and I'm glad that you found someone.”). This fast tech evolution could adversely affect loneliness, mental health and emotional well-being. The mantra of "move fast, break things" may carry significant repercussions in this context, raising the question of who truly benefits from such advancements.
As a society, we are not prepared to hit pause. And we probably shouldn’t! There are a lot of opportunities on the line. However, we do need to figure out how to walk and chew gum at the same time. We need to be advancing technology while also establishing our values and setting enforceable boundaries for socio-technical systems. We need to get pillars in place to counterbalance the commercial activity – government, academia and civil society. And we need to do it fast. We need to shepherd a conversation amongst all of us around shared values. Although some fear regulation might hinder progress, I argue if we can do this, this actually may be a path for us to move faster. Our humanity, or at least our emotional and mental state, may be counting on it.
I’ll leave you with this futuristic question (but also a really old one) in the hope that you tell me how you feel about it: In a few years, what percentage of kids do you think will say that an AI chatbot is their best friend? (and maybe go read “Klara and the Sun” by Kazuo Ishiguro in the meantime). Write to me at us@technicallyoptimistic.com.
Worth the Read
“FCC Chairwoman Rosenworcel Proposes to Restore Net Neutrality Rules” Chairwoman Rosenworcel is starting a process to attempt to reaffirm that broadband service is on par with water, power and phone service — i.e., essential. The UN, in 2021, called out the need for universal access to the internet. In this case, the FCC is attempting to establish rules in the US that prevent internet service providers from blocking legal content, throttling broadband speeds and creating fast lanes for those sites and applications which pay for access. Basically, everything on the internet should be on the same playing field.
“The U.S. Needs a Million Talents Program to Retain Technology Leadership” The United States absolutely needs to reform its immigration systems, especially with regard to technology creators. The wildest part of this article is a story of Erdal Arikan: In 2009, his breakthrough in information theory enhanced data transmission speeds significantly. Unable to secure funding and a position in the US, he returned to Turkey, later partnering with China and single-handedly catalyzing the transition from 4G to 5G. China's Huawei harnessed Arikan's insights to develop early 5G technologies, dominating the global 5G market, while the US missed out on this advancement, highlighting a lost opportunity due to lack of support for Arikan's stateside endeavors. One single visa could have changed everything.
“China Is Suffering a Brain Drain. The U.S. Isn’t Exploiting It.” That being said, the Chinese tech people are leaving China. The problem is, however, they are not going to the US.
“Meet the A.I. Jane Austen: Meta Weaves A.I. Throughout Its Apps” Meta is betting on chatbots. Soon, you’ll be able to chat with Tom Brady, Mr. Beast, Charlie D’Amelio, Snoop Dogg and others throughout Instagram, Messenger, Quest VR headsets and more. What’s the etiquette for bringing up a conversation you had with someone’s digital twin when you meet them in real life…?
“Computational Power and AI” I’m a real fan of well-put-together, 101-level briefings. This is a good one that talks about the infrastructure needed to make AI work: chips and computers plus the geopolitics, institutions, markets and politics that make it all happen.
“Europol Sought Unlimited Data Access in Online Child Sexual Abuse Regulation” EU law enforcement is pushing to be able to scan everybody’s online private messages with no boundaries. A lot of it has to do with Child Sexual Abuse Material (CSAM), but even that is tricky. Apple at one point tried to deploy software to scan for this content, but walked it back. These issues have the power to redefine what we all think of as private. And once we lose that, we may never get it back.
“Tom Hanks Says AI Version of Him Used in Dental Plan Ad Without His Consent” Spotted: another deep fake in the wild. I’m honestly surprised we haven’t seen more in the political arena as we race toward 2024.
“So Much for ‘Learn to Code’” One of the most ironic things to come out of this generative AI revolution (and I don’t mean Alanis Morissette ironic) is that software engineers helped create tools like ChatGPT and GitHub CoPilot, but it's probably now those same engineers who think that it will have the biggest impact on their livelihoods.