Media and AI
An industry that has wrestled with endless change is about to face the biggest yet.
Welcome to People vs Algorithms #52.
Well, I made it to 52, which was the goal. Let’s see where we go from here. Thanks to all for reading and encouragement. Appreciate it very much.
This week on the podcast: People vs AI.
Thanks to AI, 2023 will likely be the year when we fall in love again with the magic that can be technology and also confront the thorny questions around misinformation, the nature of ownership and copyright and, oh by the way, will those of us who have mostly typed words for a living need to be retrained to do something else? There is also the grand existential question: Is it going to replace us?. Listen here.
Indistinguishable from magic
Recent AI advancements have put humanity in a state of deep wonderment. Everybody I’ve talked to who has tried Open.AI’s new chatbot, ChatGPT, seems to react with similar astonishment. Indeed, we expect technology to do cool, useful things for us. Usually that relationship is sort of linear, understandable and mechanical—put something in, the machine does the work and gives us something back. This is different.
The internet has connected our hive mind, digitized human thought, and provided a training ground for increasingly powerful pattern-detecting models, unleashed by steady improvements in compute power. Collectively they have created an experience that mimics consciousness. AI is challenging our proprietary notions of sentience, disrupting not an industry but our conception of what it means to be human.
Perhaps, in the end, we are but sophisticated sensors feeding an organic neural processor. That biological computer, left to evolve over thousands of years, grew a sense of consciousness, seemingly divine but ultimately very machine-like. The nature and limits of consciousness seem to be slowly coming into focus.
We are evolving to a state of technological augmentation, a time when collective intelligence is measured as the sum of both biological and digital computing power. We made the tools, now the tools are us.
For years we’ve talked about AI as a powerful but invisible thing that makes other things work better. Like, how the algo made TikTok irresistible. Or, how our proprietary AI makes our ad targeting thing better, etc. But over the past year or so, we've put the technology in the hands of regular, non-technical people. If the latest AI seems toy-like, therein lies its power to capture the popular imagination and catapult a cascading wave of innovation; manifesting in surprising ways, in places we thought technology would disrupt last. Creativity is evolving from what we can make to what we can imagine. Our stories are being enhanced and extended by machines. Unexpected connections are being generated.
Barely coming to terms with two decades of wrenching change, the robots are coming to mix things up in media again.
Media and everything else. Last week's release of Open.AI's next-generation large language model in the form of a query box called ChatGPT, revealed just how powerful this stuff is quickly becoming, for applications ranging from information synthesis to code review to narrative construction. Want some examples? Start with a list of awesome ChatGPT prompts that shows its power and scope.
Or a few other things that came up in my feeds this morning:
Here's a tool to create sophisticated fictional conversational characters. This thing helps venture capitalists analyze legal contracts. For 33£, this will write your investment pitch deck. Why not make a Rick and Morty storyboard. How about an incredible selection of fictional stills from old science fiction films? Here's a new vertical search engine that returns code snippets. This guy created a Linux virtual machine inside ChatGPT.
Rumors have that OpenAi's GPT-4, expected to launch between December 2022 and February 2023, will offer another significant leap forward in model sophistication.
Media faces next wave of change
Media will change fundamentally. Let's take a moment and speculate as to how. I thought it best to start with AI. Perhaps ChatGPT could set the context and provide some practical guidance. First, on the nature of media and its importance, the response was expected and quite practical:
Media refers to various forms of communication and information dissemination, such as television, radio, newspapers, magazines, and the internet. It is a means of transmitting information, ideas, and messages to a large audience. Media plays a crucial role in providing information, entertainment, and education to the public.
Bla, bla, bla. But why is it so important? A little more nuance:
Media also plays a crucial role in shaping public opinion and influencing decision-making. Through the power of the media, individuals and organizations can reach a wide audience and make their voices heard.
Additionally, media serves as a platform for individuals and communities to connect and engage with each other, fostering a sense of belonging and community. It also helps to promote diversity and inclusivity by providing a platform for marginalized voices to be heard.
Generous. But what will AI do to media? This made some sense, but the idea that it had the potential to increase revenue seemed a bit optimistic:
AI can also be used to automate tasks such as content creation, curation, and distribution, allowing media companies to produce and distribute more content in less time. Additionally, AI-powered tools can help media companies to better target their audiences and deliver personalized content, improving user engagement and increasing revenue.
I pushed on this a bit. Remember, the AI is channeling us. And, because it's not trained (the data set is pre-September 2021) on a current conversation, perhaps it was more sanguine about its impact:
AI has the potential to be very beneficial for people working in the media industry. For example, AI can be used for tasks such as content creation, data analysis, and audience targeting, which can help media companies save time and resources. Additionally, AI can help media companies better understand and serve their audiences by providing insights and predictions based on data. However, it's important to note that AI can also pose some challenges and potential risks, such as job displacement and bias in algorithms, so it's important for media companies to carefully consider how they use AI.
Yes, AI can be used to write articles that used to be written by humans. In fact, some news organizations already use AI to generate news stories, particularly for sports and financial news. These AI-generated articles are typically based on data and statistics, rather than human-generated content, and are often used to supplement the work of human journalists. However, it's important to note that AI-generated articles are not always as nuanced or in-depth as those written by humans, so they may not be suitable for all types of news stories. Additionally, there are concerns about the potential for bias in AI-generated content, so it's important for media companies to carefully consider how they use this technology.
ChatGPT likes to find balance:
On the other hand, there are also concerns that AI could reduce the value of media by replacing human journalists and creators with automated systems. This could lead to a decrease in the quality and diversity of media content, which could in turn make it harder for media companies to attract and retain audiences. Additionally, AI could potentially lead to job displacement in the media industry, which could make it more difficult for media professionals to find work and earn a living.
Overall, it's likely that AI will have both positive and negative impacts on the media industry, and it will be up to media companies to carefully consider how they use this technology in order to maximize its benefits and minimize its potential risks.
Informative, sure. A starting point, ok. But clearly unsatisfying.
Mario Gabriele from "The Generalist" has a term I like. He calls our emerging reality "endless media," wherein "AI is not only a capable creator but an instantaneous, economical one. Over time, it may match or surpass human abilities across mediums, leading to a world in which creating a film, comic, or novel can be done on demand, ad infinitum."
This was our topic on the People vs Algorithms podcast this week. Alex Schleifer made a similar point to Mario's—AI fundamentally changes the way we think about media, from discrete artifacts that we navigate with Google, to an unending stream that is created at the end of any prompt.
The conversation had me thinking about how fundamental this change is, and how deeply it changes where value resides in the media industry. My take is it will serve to accelerate disruptive trends that the internet has had on media over the last twenty years:
Content abundance will continue to reduce the value of institutional media. In particular, the value of human-curated service and informational content will approach zero.
Highly original, deeply human perspectives will increase in value. Humans will triumph at "taste." AI will not identify the next important wave in culture.
Finding knowledge and perspective in the commons, the collective point of view of millions of contributors to a global knowledge pool will displace more and more of what has been traditionally reporting and packaging of news stories.
Aggregation of points of view under a human-led brand umbrella, like the newspaper, will have less value as intelligent agents do that work for you in a personalized manner. The value of IP, original characters, and storylines will find a way to capture the value that is ultimately remixed and extended by intelligent machines. The feed and your community replaced the front page. AI will disaggregate and hyper-personalize the feed.
AI will power limitless extensions of existing IP, extending universes and characters in ways unimaginable today. AI will generate the narrative scaffolding of the metaverse.
Yes… more, cheaper, less trustworthy, endlessly fragmented and personalized. Let’s dig down a bit on near-term consequences:
Adoption of AI in media will be rapid
Expect editorial organizations to experiment quickly. I am seeing it in companies I am involved with. First AI will provide a base for content creation as we are seeing with some of the early tools like Lex, Jasper, and Grammerly. More broadly, AI will find its way into every creative tool. Content creators will sit on top of these queries, bridging the gap between what is happening in our world, and what the SEO gods are asking for spitting out ever most sophisticated queries to generate the first draft of stories.
As AI becomes more sophisticated it will slowly displace more of the editorial effort. AI ambition will steadily move upstream as models mature, imitating human point-of-view sourced from a stream of real-time chatter, aggregating and presenting that content to meet the hyper-personalized requirements of the reader.
Enterprising editors will survive as AI-powered human hybrids. More will be commoditized. Copy editing is surely a dying vocation. The existing model of Google-led SEO / SEM Google will be suffocated under the tsunami. Monetization will struggle with abundance.
Human IP will power personality, tone, and style
Just as generative art can ape the style of Dali or Wes Anderson, AI will present content in the voice of brands or distinguished individuals. Slowly these will come to not just represent tone but distinct points of view, influencing how information is curated and personalized; “write a review of the iPhone in the style of Walt Mossberg.”
On the podcast, Alex revealed he has the computer read news in the voice of Snoop Dog. Interesting choice. This is basic voice IP exploitation and is just the beginning. Like a good personal editor, expect AI to automate sourcing, synthesis, and presentation of content in increasingly bespoke ways.
Google faces existential challenges, Twitter and Reddit become more valuable
No Google is not going away tomorrow. Yes, they too are investing heavily in this stuff. Of course, there are many, many practical limitations that will take time to unfold. But, now we can see a new path.
A wave of competitors will be incented to put more value in the query response, bypassing the link. Google has carefully managed its position as a weigh-finder and question-answerer. This balance is, of course, necessary to persist the best media monetization model ever known and Google's delicate monopolistic position as the backbone of the open web. Their position will slowly deteriorate as a new model emerges that smartly generate responses with a combination of text, image, video, and VR, created in real-time to the sophisticated dictates of the user. The idea of a link naturally loses value.
We are seeing the limits of search today. A common practice to avoid a gamed list of Google query responses is the point the search engine at a conversational data set like Reddit with the addition of ":Reddit" at the end of a query. More of our knowledge will bypass links in favor of base-level knowledge, aggregated in narrative packets.
Change will come slowly. Currently, ChatGPT is trained on a pre-September 2021 data set. It will have to become real-time. Mechanisms to validate AI accuracy are not yet contemplated. The computational requirements to scale something like ChatGPT to the global volume of Google are overwhelming. Insiders suggest a GPT-like experience would require 10-100x the computing power to support current search volume. Expect Google’s product to evolve more rapidly as these concepts popularize and costs fall.
But, you will see a proliferation of niche search/AI entrants, especially against narrow use cases.
Finally, a very good point was made by Alex this week; the Twitter data set will grow in value as a gathering point for millions of points-of-view and input to the AI synthesis engine that can make sense of it.
Verified by a human, made by AI
Like the design attribution made famous by Apple, "Designed in California, made in China," the avalanche of AI-generated content will demand quality signifiers that create trust. AI will leave you wondering if what is written has been verified by a human that knows what she/he is talking about. Media brands will play role in establishing credibility. New signifiers will emerge to guarantee the veracity of content.
AI models will be the next big "media" brands as will proprietary data sets. They will be aggressively sought out as the underlying drivers of value.
Old aggregation is dying
Traditionally media played an important role as an aggregator of topics and points of view. The front page presented the day's most important stories, magazines profiled notable personalities on covers, etc. Algorithms driven by the crowd have steadily replaced this aggregation function, as have pointers from influential personalities on social media.
The role of aggregation will continue to shift with AI as legions of smart assistants dutifully sort, gather and present information according to preference and consumption history.
The effect will be to continue to erode the media brand as a cultural arbiter. It will exaggerate media fragmentation and a landscape of fractional truths.
The blockchain will play an important role in verifying legitimacy
Fred Wilson makes a good point here:
I think AI and Web3 are two sides of the same coin. As machines increasingly do the work that humans used to do, we will need tools to manage our identity and our humanity. Web3 is producing those tools and some of us are already using them to write, tweet/cast, make and collect art, and do a host of other things that machines can also do. Web3 will be the human place to do these things when machines start corrupting the traditional places we do/did these things.
Advertising becomes AI-driven selling
Expect AI to force continued commoditization of much of what the ad industry is paid to do. But a more profound change is in play. AI will turn static marketing funnels into personalized sales experiences. It's not difficult to see how a well-trained bot will connect a positioning headline with a back-and-forth sales experience that bridges brand advertising, media, merchandising and service.
That’s enough for now. Lots to process. Let’s end with this: AI, play me an old-fashioned rock song, one that summarizes this long post, something that takes me back to 80’s, top 40 radio in Regina, Saskatchewan. Hold on loosely…
Have a great weekend…/ Troy