The AI Race
There is a war going on, who beats who in the AI race. Companies are fighting between themselves on what each of them can bring to the table and who can bring them first. Currently the debate is between OpenAI, Google, Microsoft and maybe Apple. Here are some of my thoughts on what’s happening and where things could lead to.
The magic is over now, people who what is Generative AI and how its built, its no more a fantasy tool. Just like NPM Packages we’ve LLM models floating across the hugging face space. And everyday they get better and better, in-fact most of the models are open source thanks to Meta for Llama and Mistral for their suite of models.
Now its time for the market to consolidate and sit in their roles and extend the growth of LLMs in the future. And I see every company here has a role to play and they are unfortunately limited to their roles due to their current placement in the market.
OpenAI - The pioneer
Even though Googlers were the one who published the patent for transformers model, OpenAI was the one who put it to the hands of people (considerably a larger crowd). They have been doing this for quite sometime and very well known by people. Their models and product like GPT, ChatGPT have become the acronym of Generative AI (Like Xerox for photocopy).
They are the pioneers of LLMs and their applications, but they have a very serious limitation to grow, which is the ecosystem. I strongly feel they could be a really powerful model company who build the crazy models for the people know what is possible. But putting it to everyday tools and systems is a challenge for them until they have a loyal partner (Like Microsoft).
Microsoft - The recent resurrect
People who use internet knows Microsoft. They have been in the market for an eternity now, one or the other we all have used at least one of their product in our lives. But with their dip in the market and their come back as a giant changed the landscape for them. They lost the consumer market while racing the enterprise. But luckily it worked out for them. They are almost everywhere in enterprise, and this made their game easier to compete in the AI race.
Just like their proposal to build operating system for IBMtheir investment in OpenAI in early states of the model company gave them a huge leverage in the race. They we’re the first to launch the code assistant called [Copilot in Github] (one of their acquired product) which blew the market and created an impact is most developers. With the success in that they got their hold in the race and started pumping out AI services in Azure (their cloud platform). But these services were just a wrapper for OpenAI with added security and compliances. Following to that was a master stroke of implementing Copilot for Microsoft office. And they we’re able to poach the enterprise customers of OpenAI with these services. And I strongly feel they will float around the enterprise segment for a considerable future.
Google - The deep tech company
Recently it has become the most controversial company when it comes to generative AI game. Inspite of being the owners of the transformers white paper, they didn’t focus much on the generative AI space. This made them a slow mover in the game. In order to compete in the game, they made some rash decision and half backed solution in the early state of the race mainly when they launched the Bard. This costed them money, reputation and customer distrust.
But if we think that these would push the back, we’ll be wrong. Google is the only company that has the muscle and the ecosystem to push Gen AI like nobody else. If you have the last keynote (2024) you could see them already integrating Gemini (their flagship mode) into almost all of their products.
Unlike Microsoft I think Google will float around the consumer market segment where they have put the LLMs to use in the hands of general public in their wide range of Products.
Apple - Privacy King.
This company will be the slowest of all, yet they will bring out some ground breaking technology and also really well thought use cases for generative AI. They will stick to personal computing and maybe into some of their media tools like Final Cut pro, but they will try to sit to on device LLMs. This will open up a new ways to optimise other models to run on low powered devices yet being powerful.
I don’t think Apple will try to enter either enterprise or the cloud models space, which is a good thing for privacy first people.
I cant wait for them to make Siri as LLM Powered
Meta - The new nice guy.
Since the time of they privacy law suits, Meta has been waiting for a chance to enter into the good books of people and now they got one. Open sourcing their powerful models (Llama 1, 2 & 3) back and forth, they gained some trust in the developer community and also helped them to erase old name of theirs from people’s mind.
Its better if they float around the open source way because that keeps the race more neutral and open for many others like Mistral to come in. And Mark has promised us that they will be good when they reach AGI
Who will succeed with AGI?
Honestly this is a question that even I’m trying to find answers for, and I feel it will be a joined effort by two companies. Maybe Google and Microsoft, or Google and Apple. But who ever gets to AGI first, it will change the scene in the valley forever.