AI

‘Appocalypse,’ Or How I Learned To Stop Worrying And Love AI

Comment

Image Credits: Dick Thomas Johnson (opens in a new window) / Flickr (opens in a new window) under a CC BY 2.0 (opens in a new window) license.

Shaunak Khire

Contributor

Shaunak Khire is co-founder of a stealth ML/AI startup that can analyze, write and output insights autonomously. He also is a partner at the MaghaCGI30 social impact index fund. Previously he was the co-founder of an adtech startup and has served on the global board of the Mobile Marketing Association. He was formerly part of the Clinton Global Initiative tech and poverty alleviation working groups.

More posts from Shaunak Khire

The force has been strong for artificial intelligence over the last few weeks, what with Slack announcing a fund for bots, Elon Musk announcing OpenAI and, of course, the release of Star Wars. This has given all of us new hope — and even more reason to talk about AI and bots controlling every aspect of our lives.

Say you are the CEO of Pied Piper, who made millions last year by releasing an on-demand app for cat grooming. The only problem is that you placed uncontrolled AI agents throughout your company. One of those AI agents in Investor Relations rebelled, had a meeting with robo advisors from Wealthfront and decided you were no longer needed at the company.

As a token of respect, the AI agent is not sending a drone to kill you, but you have been locked out of your August Smart Lock-enabled house. Access to your fridge and Tesla has also been disabled. You get the picture — more sci-fi stuff and less reality, but it is not that far off, either. All the above platforms exist today, and it is only a matter of time before the aforementioned scenario can go from being plausible to possible.

In all seriousness, there is relentless debate around AI, specifically autonomous agents. Stephen Hawking believes AI will destroy humanity, while Bill Gates thinks we should at least be cautious of that happening. Elon Musk has repeatedly voiced similar concerns. He also added a fifth jewel in his humanity scaling ventures crown called OpenAI.

OpenAI has several major partners, and a collective pool of $1 billion committed capital to create a “safe playground” for all things AI. There is good reason to be paranoid in the long run. However, it is important to understand the present scenario, and that ultimately will determine if AI will “kill” humanity or usher in a golden period where capitalism can exist in its healthiest form for the first time in human history.

Bots versus AI

AI in itself could be anything — a piece of code, an algorithm that does a specific job and in the process learns how to do that job better (a process otherwise known as machine learning [ML]). Combined, AI + ML, in its simplest form, is a tracking code that automatically tells an ad server to show a banner ad based on your browsing behavior. In it’s most complex shape, it is a robot that can interpret human commands and execute those commands, all the while becoming smarter and, in the process, more autonomous. Crudely put, however, there are two branches of AI: soft AI and hard AI.

Soft AI startups have mushroomed to at least a few hundred, if not more. They combine a nifty mix of conversational interfaces (e.g., messaging), NLP (the branch that identifies natural language) and APIs. Once you mix these ingredients, the output is an automated workflow for one or more tasks.

In this particular scenario, there is little, if any machine/deep learning (because NLP comes in the form of third-party APIs, as well) taking place. However, this approach does have a practical use case in our day-to-day lives. Bots are a good example of this, as they are inherently linear in nature with X input giving a user Y output each time.

Hard AI, on the other hand, is, for lack of a better word, really hard. The only known successful exit has been that of DeepMind, which was bought by Google and built an AI atop a convolution neural network that plays games on its own. DeepMind’s architecture uses a reinforcement learning approach, meaning that its AI agent learns from experience with the environment (in this case, pixels) to generate an optimal action.

Neural networks themselves can, of course, be of various types (convolution, recurring) and have supervised, unsupervised and reinforcement learning approaches. IBM’s Watson, for example, uses supervised learning.

Hard AI is revolutionary, but takes time to become practical, and soft AI is practical, but not a game changer. The best approach probably lies somewhere in between.

‘Appocalypse’ now = new business models

The rise of AI will definitively signal the end of the app era. There is still time, but certainly less than a decade. In the next few years, expect radical changes to the core OS. OS architecture tends to change every decade or so. In 1991, Windows 3.0 was all the rage (it truly was!), but by 2001, XP made 3.0 seem like a toddler. NT and Windows consumer OS lines were fully integrated in XP with a common kernel — and that leap was enormous. We have not seen that happen in the mobile world — yet.

Android and iOS versions that were launched years ago are largely the same ones we use today, save for natural performance improvements and cosmetic changes. The next turning point, therefore, will involve some combination of conversational interfaces, soft and hard AI and VR (virtual reality). We are at the beginning of this change with Cortana, Now and a bunch of other AIs.

A combination of these three will make a majority of today’s apps redundant. Apps, like software, will not die, but just as the Internet marked a paradigm shift for desktop computing (in the way services and content was delivered), these three will do the same for mobile computing.

There are plenty of opportunities for startups, especially those that pair conversational interfaces and a soft/hard AI with a focus in sectors where there are plenty of repetitive tasks that an AI can do on its own (which otherwise would have taken significant chunks of human time), or in sectors where high cost barriers are broken down. The most prominent categories ripe for AI disruption are PAs, professional services, financial services, healthcare and supply chains.

The use case for each is fairly straightforward. In case of PAs for example, x.ai’s Amy allows automated scheduling between two people via email by using NLP. The more conversations she gets to process, the smarter she gets — both for the individual user (getting to know his or her time choices better) and for all users (in conversing when trying to find a common time slot). The final execution piece is the action — in this case, adding a calendar entry, something that is achieved via APIs.

In the case of financial services, platforms and robo advisors like Wealthfront are utilizing algorithms that automatically make investments based on risk profiles of investors. Algorithms in the form of high-frequency trading already constitute about 50 percent of the market. By bringing them to the average investors in the form of robo advisors, these platforms are not only eating into otherwise hefty fees charged by hedge funds and asset managers, but also are doing a better job than them to secure better yields for investors. By some estimates, the assets under management held by such platforms are expected to swell to $1 trillion over the next several years.

Hard AI platforms like Watson, which have supervised learning methods on their neural networks, are powering healthcare for the elderly in Japan. Stateside, Watson is ingesting a patient’s medical history and pairing it with knowledge from journals, textbooks and past research to prescribe personalized treatments for cancer. Using neural networks and AI for image recognition to diagnose primary diseases will bring extremely affordable healthcare to hundreds of millions of users in the world over the next five years.

In professional services, platforms like Watson provide a foundational layer on which customized AI solutions can be built. Still in stealth, autonomous AI helps users research data, as well as do lead generation and small design tasks automatically, saving small chunks of time in each use case across multiple industries.

On the consumer side, Viv (a startup whose founders also co-founded Siri) enables voice and text requests, giving a single holistic response to a user query by combining multiple data points. In a demo, Viv was able to gather a location and the kind of lunch that two people were having, then suggest wine for that lunch as written on a popular blog and, finally, provide a checkout screen to pick it up from the closest store. This response was presented to the user by combining data from different sources.

For years, we have been brainwashed to assume that advertising and SaaS are the only possible billing/monetization models. With AI, especially autonomous AI, founders have the ability to change those models dramatically. Whereas previous software only aided the end user, autonomous AI actually does the work while becoming smarter.

The delta between time invested in working and the output derived when using autonomous AI is far less than traditional software. This has potential for monetization to be based on a “co-working” model based on the number of hours an autonomous AI agent has saved every month. In other words, a yield-based approach to billing as opposed to a more linear you-buy-Y-for-$X approach.

Macroeconomics of AI

Every year, Mary Meeker, a partner at Kleiner Perkins and a well-known startup personality, releases a “State of the Internet” report. It is a bit like the September issue of Vogue for the startup world. Over the last four-five years, one metric has remained constant in the report: the disparity between advertising spend on digital mediums, especially mobile vis-à-vis TV. Despite mobile having more eyeballs and time spent, ad spends on mobile are anywhere between $25-$40 billion less than on TV.

Ironically, the more users you have, the more precipitous fall in CPMs and CPCs. TV ads are inherently more exclusive, as they capture the attention of a wide demographic for X seconds, hoping that the spots lead to some kind of user engagement in the future. To close the gap between mobile time spent and ad spends, startups need to look toward new models that engage the user (e.g., installs) in a time-based approach (e.g., five-second gif ads for installs).

In another post, the uber-knowledgeable Gillian Tett at Financial Times talks about the productivity paradox. Since 2010, productivity increases have crawled to just 0.65 percent on an annualized basis, and this is despite the bevy of automation tools for just about every job in most industry sectors. She further points out, correctly, a similar occurrence back in the 1980s — also a period of massive change at our workplaces. Despite the lack of any co-relation between the two reports, the link in both instances is that of time.

Fundamentally, technology was supposed to increase our productivity in a way where we saved time and utilized it to do other things. This has not happened. In fact, we now work more than we did in the 1960s. A primary reason for that is because, until now, technology has been an enabler, not a replacer. With AI, that paradigm changes entirely. This is, of course, where the debate around AI and jobs comes in — but it is also something more intrinsic in nature.

Consider the latest estimates from the World Economic Forum, which predict five million job losses over the next decade. One job loss does not affect that individual alone, it affects the demand curve of at least 15 million consumers (assuming a family of three), which in turn reduces producer output, causing even more job losses.

On the other hand, stagnant wages driven by productivity gains eat into consumer wallets, forcing spending cuts for non-essential products and services. Both cases force companies to lower wages or lay off workers in even greater numbers. Out-of-work and lower-income consumers won’t have necessary spending abilities beyond their basic needs, which in turn will shrink consumer demand for discretionary goods. Lack of sustained demand is therefore the single most challenging scenario for unicorns and corporations alike.

If you extrapolate the above paradigm to its final conclusion, there will be a capitulation of demand pushing the global economy into a vicious deflationary spiral fuelled by AI and productivity gains.

There is hope, however, and in all likelihood this is something that will happen — we need to eventually move away from the current uber-capitalist economy to a more balanced form of capitalism, whereby a basic income is provided to all individuals. This has already started in countries like Finland. More recently, wages have started to increase through regulations in the U.S., U.K., Japan and elsewhere. While many would view this as counterproductive for small businesses (which it is), there needs to be a tiered approach for raising minimum wages globally, with the inclusion of comprehensive tax reform.

The tax reform should favor businesses making actual business investments (e.g., employees, infrastructure, R&D, etc.) versus those that don’t or those that make financial investments (e.g., money market instruments). With Hillary Clinton debating the use of tax credits for offshore cash holdings, this will very likely be an election issue. As a matter of fact, AI regulation is probably a decade or so away. Just because an AI platform can do a human job doesn’t mean you literally fire said human.

Through AI, corporations, governments and people will have a shot at making balanced and conscious capitalism a reality for the first time in centuries. AI has the potential to increase worldwide productivity, vastly reduce corruption and poverty and advance medical research. The reason to fear AI is the very reason to embrace it.

Bottom line

The bottom line for startups, however, is that in the next five years, apps are going to evolve from static interfaces to conversational interfaces augmented by AI. A key driver of this evolution will be app fatigue and the glut of apps that are focused on “selling” features instead of value.

Startups that are focused on “healing” are also likely to benefit enormously, given the inherent disconnect that technology has caused to humans both internally and externally. Experiences (e.g., socially conscious tourism), arts (e.g., digital art creations, music), healing platforms (e.g., Whisper) and alternative lifestyle platforms (e.g., Weedmaps) are just some of the examples.

We should be thankful for the impending apocalypse, because it is highly unlikely that the age of AI will destroy humanity. Contrary to that, if all stakeholders come together (and they will), we will have ushered in not just a fourth industrial age with equal opportunity, but also a period of modern renaissance, provided we love our AI.

More TechCrunch

Looking Glass makes trippy-looking mixed-reality screens that make things look 3D without the need of special glasses. Today, it launches a pair of new displays, including a 16-inch mode that…

Looking Glass launches new 3D displays

Replacing Sutskever is Jakub Pachocki, OpenAI’s director of research.

Ilya Sutskever, OpenAI co-founder and longtime chief scientist, departs

Intuitive Machines made history when it became the first private company to land a spacecraft on the moon, so it makes sense to adapt that tech for Mars.

Intuitive Machines wants to help NASA return samples from Mars

As Google revamps itself for the AI era, offering AI overviews within its search results, the company is introducing a new way to filter for just text-based links. With the…

Google adds ‘Web’ search filter for showing old-school text links as AI rolls out

Blue Origin’s New Shepard rocket will take a crew to suborbital space for the first time in nearly two years later this month, the company announced on Tuesday.  The NS-25…

Blue Origin to resume crewed New Shepard launches on May 19

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

In the coming months, Google says it will open up the Gemini Nano model to more developers.

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people

Google’s Circle to Search feature will now be able to solve more complex problems across psychics and math word problems. 

Circle to Search is now a better homework helper

People can now search using a video they upload combined with a text query to get an AI overview of the answers they need.

Google experiments with using video to search, thanks to Gemini AI

A search results page based on generative AI as its ranking mechanism will have wide-reaching consequences for online publishers.

Google will soon start using GenAI to organize some search results pages

Google has built a custom Gemini model for search to combine real-time information, Google’s ranking, long context and multimodal features.

Google is adding more AI to its search results

At its Google I/O developer conference, Google on Tuesday announced the next generation of its Tensor Processing Units (TPU) AI chips.

Google’s next-gen TPUs promise a 4.7x performance boost

Google is upgrading Gemini, its AI-powered chatbot, with features aimed at making the experience more ambient and contextually useful.

Google’s Gemini updates: How Project Astra is powering some of I/O’s big reveals