Not really if you have to divide resources between two. Computing necessary for training these models is not cheap and there is an obvious opportunity cost here.
I have no idea what MS is doing with AI internally, but predictive text is only one of the avenues towards AGI which sure seems to be the direction OpenAI (and everyone else currently looking to sell a product) are going. There are certainly other directions MS can go in the same field without putting all their eggs in one basket.
Back both horses and drop the one that loses the race.
The thing is that there many many horses in this race
Having two entrants is still better than one.
Not really if you have to divide resources between two. Computing necessary for training these models is not cheap and there is an obvious opportunity cost here.
MS has more than enough cash and resources to back two horses.
This also gives them the luxury of trying a different approach.
Time will tell my guess is that Microsoft will sale its stake in openai or just drain openai from resources and people until it will disappear.
I have no idea what MS is doing with AI internally, but predictive text is only one of the avenues towards AGI which sure seems to be the direction OpenAI (and everyone else currently looking to sell a product) are going. There are certainly other directions MS can go in the same field without putting all their eggs in one basket.