Receive free Artificial intelligence updates
We’ll send you a myFT Daily Digest email rounding up the latest Artificial intelligence news every morning.
In May 2022, Morgan Stanley published a big report on the global side-hustle economy. It mentions crypto 40 times, blockchain 22 times, decentralised autonomous organisations 33 times, play-to-earn gaming 27 times and artificial intelligence just once, in passing, in a list of themes that wouldn’t be mentioned in the report.
Today, Morgan Stanley publishes an update that hardly mentions crypto and has a lot more AI — because AI’s given as the main reason its first report was too cautious.
Contrary to the belief that gig working and employee bargaining power would diminish post-Covid, multi-earning has evolved into a secular growth theme, with Generative AI the differentiator. Our AlphaWise survey shows gig-earners’ income boosted 21 per cent by using Generative AI tools versus those not.
The worlds sketched out by both notes are similar. Housing nearly everywhere is unaffordable on salaries alone, collective bargaining power is making a (minor) comeback and trust is falling in our ever-ageing business leaders. Low-growth economies and youth unemployment leave no option for many people but to hustle, while for others the pandemic revealed the value of flexibility.
Surprises since 2022 are that workers have been able to resist demands to return to the corporate panopticon — office attendance in the US still 50 per cent below pre-Covid levels and falling — and that interest in multi-jobbing has remained even as economies slow.
Gig work therefore looks to be economically countercyclical, Morgan Stanley says. Though official figures suggest about one-in-20 people globally are working multiple jobs to make ends meet, the analysts estimate that the true number of multi-earners is probably closer to one in 10.
In 2022, the bank’s surveys of work habits were all about Uber driving, day trading, NFT staking and Axie Infinity husbandry. The latest includes questions about generative AI tools, as well as widening the sample to include India, which changes the outcomes quite a bit:
Eighty percent of respondents in India claim to have used generative AI tools of any kind (vs 40-50% in the UK and US, respectively). Indian respondents claimed 63% were using Generative AI at work . . . Perhaps most the surprising result of all from our survey was that nearly two-thirds of Indian respondents believed they have a more secure full-time job as a result of productivity gains from Gen AI tools, running counter to the conventional perception in the equity markets.
A headline finding is that gig workers everywhere claim they’re making money from AI:
How? Either by feeding the internet’s main content mills with simulacrum, or by automating their drop-shipping operations:
Among the giggers, enthusiasm around generative AI seems to exceed even crypto circa 2022. Respondents whose employers have banned gen-AI are about twice as likely to want to quit immediately, the survey finds — which has an echo of last year’s result showing the high percentages of gen-Z’s wanting to leave full-time work and turn pro on Robinhood, OpenSea or DeFi Kingdom.
A problem with surveys is that people lie, intentionally or otherwise. In this case, there’s probably a tendency among respondents to overestimate how much extra they make by having ChatGPT as a wingman, or to exaggerate their tech savvy. Also factor in that Morgan Stanley’s estimates for the size of the gig economy draw on some imperfect sources, such as Google search volume for “how to make money on OnlyFans”:
A lack of hard data on the value of AI-assisted hustling means the research is sometimes difficult to tell apart from all those spammy blue-tick X accounts.
For example, here’s Morgan Stanley’s bridge chart showing how a person might build a “comprehensive suite of productivity-enhancing tools” meant to “[boost] productivity as much as 55 per cent” — all for just $160 a month.
(The up-to-55 per cent productivity boost cited appears to refer only to Copilot coding, with the same figure appearing in Github’s marketing. A $160 monthly AI subscription cost “clearly does not cover salaries, overheads or other costs associated with starting or scaling a business,” the note adds, helpfully.)
In the “real-world examples” section are just three examples: a YouTube tutorial on how to make an app with Github Copilot and Midjourney, along with these two tweets:
I gave GPT-4 a budget of $100 and told it to make as much money as possible
I’m acting as its human liaison, buying anything it says to
Follow along 👀
— Leonardo Sousa (@leo_rsousa_) March 20, 2023
GPT-4 can take a picture of napkin mockup as an input and output a fully functional website (HTML/CSS/JS) 🤯🤯🤯 pic.twitter.com/fIHB12RTHp
— kitze 🚀 (@thekitze) March 14, 2023
Much more expansive is a section on how a person might seek to monetise their digital mulch. They could make a Shopify shopfront, maybe? Or start a blog? Or put things on TikTok, Instagram Stories or Twitch to harvest the ad dollars? And have you ever heard of X?
As it relates to generative AI, X is now paying smaller creators for the content they produce on the platform, and the figures for creating impressions are larger than many had thought. This is encouraging more creators to use generative AI to make sufficient content to reach monetisation thresholds. Elon Musk recently noted that what matters is how many ads were shown to other verified users. For example, X creators have noted strong returns by using Generative AI to substantially lift earnings. As one user stated via X, “$8 to make $370 is a fantastic ROI for something I do anyway.”
Wait! There’s more! How about taking a punt at systematic trading? Or pretending to be a tutor? Or maybe someone could become a celebrity and then deepfake themselves? These are all actual examples to be given blurbs, eg:
Popular influencers on major social media channels have turned to cloning themselves with monetisation by the minute for people wanting to engage with their likeness. This type of incremental multi-earning is undoubtedly not for the more informal multi-earning population, but it demonstrates the ways in which AI is being used to create legitimate ‘deepfakes’, although there is also the potential for monetisation via illegitimate deepfakes.
Putting an aggregate value on all this mechanised plagiarism is tricky, but Morgan Stanley has a(nother) go.
Our relatively conservative base case assumptions place global multi-earning at $250bn, with scope to nearly double to >$400bn by 2030. The largest individual difference between the maths in this report and our prior multi-earning iterations comes through the Generative AI uplift component by 2030 [ . . . ] adding 21 per cent, or $83bn, to our multi-earning income stream calculations by the end of the decade. Our bull case is $1.4tn and $300bn respectively.
— Deep-diving the disparate, desperate, PG-rated side hustle universe (FTAV)