The Greatest AI Magic Show on Earth is happening at breakneck speed
There are already tens of thousands of AI apps with hundreds more being launched daily, almost all of which seem incredibly useful, and some of which are downright astonishing.
AI is driving me crazy. There is too much going on. It was one thing trying to keep up with the internet in the 2000s. And then it got worse with smartphones and apps in the 2010s. And then crypto just about extinguished all other tech news with its metastatic growth, still ongoing.
But AI? It makes me want to go and live in a windy shack in the Welsh highlands. With no electricity and perhaps a sheep or two and the complete works of the 19th-century Russian novelists for company. I can’t keep up with AI news. It’s making me unhappy.
I sat down to write this column to summarise the burgeoning AI third-party app offerings in AI since ChatGPT was announced in November 2022. I’m afraid that the task has bested me.
For instance, entirely by accident I came upon an AI app that promised to build beautiful and impactful Powerpoint presentations. So I gave it a test job to build a presentation about a complex, difficult, multilayered, politically charged and philosophically fraught technology that I had recently written about. I requested 20 slides. Asked it to include some really arcane stuff, and leave out other stuff, and to give case studies and examples and anecdotes to engage an audience. I asked it using five sentences of vernacular English, hastily written.
It asked me sweetly whether I wanted to upload a corporate logo. It offered me some design ideas, all rather nice. And then I hit the “generate” button, expecting it to think for a few minutes and upload a presentation which would need lots of work.
But no, it built the presentation in real time, in front of my eyes, slide by slide. It was excellent and relevant. Perfectly structured. Accurate. With appropriate graphics and images to spice it up. If I was going to actually use this (I’m not), it would have taken me perhaps one hour to buff it up.
I estimate it would have taken me about 10 to 20 hours to build the presentation from scratch, with lots of hard research. That’s a 10-20x productivity gain.
I have a friend who is a well-known and deeply experienced educator in the art of photography, teaching twenty-somethings. We are all f*cked, he tells me. Actually his words were: “AI is creating a massive body blow to the simple task of teaching,” but that was the underlying viscera of his message.
He is wrestling with whether to mark students up or down for using new AI tools to edit and invisibly improve their photographs. Is it cheating or a demonstration of tool mastery? And he is a technophile, not a luddite.
And then there is the app that summarises a boring long Zoom meeting into five-minute summaries of perfectly composed narrative. It actually sheds the noise and dross and grandstanding and redundant comments and bad jokes and keeps the important stuff that needs to be remembered. For anyone whose attention wanders during video conferences, like me, this is manna from heaven.
Here’s the problem – there are already tens of thousands of these AI apps with hundreds more being launched daily, almost all of which seem incredibly useful, and some of which are downright astonishing. In every imaginable field of human endeavour from writing to medicine to science to music composition to coding to finance to business operations to religion.
Tracking sites like Bensbites.co and Turingpost.com do a brave job of trying to pick and choose the most important ones every day, but I suspect they too are looking for a windy shack in the Welsh highlands. Maybe they could be my neighbours.
It is part of my job to be on top of this stuff. With a great deal of effort it is indeed possible to follow, track, analyse and chew over the “big” announcements, the major advances, the stream of new deep-learning models – OpenAI’s GTPx, Google’s Palm 2, Anthropic’s Claude, IBM’s DeepMind, Meta’s LLaMA, Adobe’s Firefly, Technology Innovation Institute’s Falcon LLM and others.
But what has happened is that these big companies have (mostly) opened their core AI plumbing to outside access, in some cases without charge. And an army of innovators and dreamers and entrepreneurs have rushed in saying “I’ve got a good idea! I’ve got a better idea! I’ve got an even better idea!”. And they are. At least all of the ones I have seen (hundreds, thousands, I have lost track).
And here lies the substance of this column.
In 30 years of following new technology, nothing has come close to this as pure free fuel for an endless combustion of new innovations. In some cases, you don’t even need to know how to code (AI takes care of that as a side benefit) to build an app. Not even the explosion of Web 2.0 and mobile apps over the past 20 years comes close.
And this seems different. In its very brief life, bot-driven generative AI has quickly become a fecund, fast-growing and tangled undergrowth where it is hard to see what is really going on, what will flower and what will die on the vine.
Consider this screengrab, taken from the daily AI newsletter www.bensbites.co. And this is certainly just a small single-day sampling of the stream of stuff coming out of the AI hosepipe.
When I look at the volume of new AI applications being offered to a somewhat bewildered public, I wonder if this is just a repeat of the early years of the internet where thousands of newly created static websites competed for attention and then were eventually abandoned to live out lonely lives in the unvisited graveyards of dead domain names.
So, what really is the difference with AI apps? The key is AI’s ability to understand and communicate (both as listener and talker) in common human language (keep in mind that Meta’s recently announced Massively Multilingual Speech AI research models can identify more than 4,000 spoken languages, 40 times more than any known previous technology). That is not only new. It is a holy grail of sorts. It has bridged the history-long gap between human inspiration and technological capability.
So, what is a befuddled technology correspondent to do? Decamp to my mythological shack in Wales? Or take a ringside seat to the magic show currently on display, eating popcorn as I stare in drooly amazement?
Or perhaps grimly to soldier on, trying to make sense of it all. DM
Steven Boykey Sidley is a Professor of Practice at JBS, University of Johannesburg.