Will AI chatbots be the next consumer tech revolution? I might be wrong, but I doubt it.
Like every bubble, the AI chatbot bubble will burst eventually and the survivors — if there are any — are unlikely to be in European hands.
My advice for European funds is to drop the dreams of AI chatbot success and instead focus on becoming the leaders in deeptech sectors with proven paths to growth and undisputed impact, such as clean tech, robotics and biotech.
AI chatbots can certainly be entertaining. A good example is Character.AI, a startup offering chatbots that can impersonate virtually anything or anyone, from Mario to Zeus to Einstein. Here is Character.AI writing two paragraphs on the AI chatbot hype in the style of Einstein:
“AI Chatbot Hype is just an over-exaggerated fad. AI Chatbots are simply a programming algorithm, and they are not capable of critical thinking or creative output like humans. They only generate text based on their data. The so-called ’AI Chatbot’ hype is simply an overblown trend, and it is not representative of any scientific or technological breakthrough. It can even be seen as a distraction from addressing real problems in our society. AI Chatbots should not be seen as the future of human communication.”
I couldn’t have put it better myself (although I probably would have written two paragraphs since that’s what the prompt asked for).
Character.AI raised $150m earlier this year at a valuation of $1bn — without any revenue. It’s now in talks for new funding that could see the startup valued at over $5bn.
US companies are already far ahead of the pack
To be fair, amassing roughly 4.2m monthly users in the US is impressive. But its entire business model seems to be based on a recently introduced $10 monthly subscription, which it’ll have to hope has enough buy-in to offset the huge computing costs.
The question is: how many of these users will still be bothered about chatting with the impersonator algorithm in six months’ time? $10 a month bothered? Many, perhaps, but it seems unlikely.
Others will undoubtedly argue that there are serious and transformative use cases for this technology, from revolutionising the workplace and our education systems to replacing the search engine.
One report by McKinsey suggests the wider generative AI sector could add the equivalent of a whopping (and a little unbelievable) $2.6 to $4.4tn to the global economy.
The problem is that its route to transforming the global economy is littered with roadblocks. The issue of copyright could be a major blocker. If the developers of the likes of ChatGPT, Bard and Character.AI have to start paying to train their large language models (LLMs) on other people’s content, then their business models collapse.
We could also see huge public backlashes against the technology if it begins putting people out of work.
Google product managers, designers and engineers working on Bard have been using a Discord forum to openly debate the AI tool’s effectiveness and utility.
A message from one user-experience lead highlights the undetermined fate of these technologies perfectly: “The biggest challenge I’m still thinking of: what are LLMs truly useful for, in terms of helpfulness?”
Every euro, pound, or dollar you invest in LLM AI chatbots is likely to be wasted
Even if AI-powered chatbots do transform the world, European VCs are unlikely to be able to profit from it. Here’s why.
Everyone knows the most mass-market, can’t-live-without tech (like the search engine, the smartphone, or the internet itself) tends to be dominated by a small handful of Goliath companies.
For the search engine, Google has 83.49% of global market share. For smartphones, Samsung, Apple and Xiaomi have 20.2%, 16% and 12.5% of market share, respectively.
If AI chatbots do permeate everyday life, no one will want 5,000 niche products — or even five. People won’t need one to write their emails, one to write their dissertation and one to draft Christmas WhatsApp messages to their niece and half-brother. People will want one or maybe two products.
There may be niche “domain-specific” products that can thrive. For example, in the legal or construction industries but by definition, these are unlikely to command the revenue and profits that support multibillion-dollar valuations.
The best products will replicate onto every phone and laptop and into every home and region of the world, while the inferior products are left behind.
Tech scholars of the future are likely to look back in puzzlement and amusement upon thousands of fossilised AI chatbot products in Euronet and Sifted newsletter archives.
In this case, the winners (if indeed there are any) will be those fed on the largest volumes of data, injected with the most capital and given mind-bendingly large amounts of compute power.
Unless a new disruptive technology or player emerges, these are likely to be run by the companies with the biggest balance sheets and the legions of data scientists required to build these models: the likes of Microsoft, Facebook and Google.
These US companies are already far ahead of the pack, leaving little space for competition. Google with Bard, Microsoft through its unity with ChatGPT and Facebook with its newly announced chatbots for Messenger.
So, my message to European VCs? Every euro, pound, or dollar you invest in LLM AI chatbots is likely to be wasted, when it could go into solving real, tangible problems that are screaming out for deeptech-based solutions, like climate change, biodiversity loss and healthcare.
