How to invest in AI?

isOur this years first panel discussion about AI investing was organized online via ZOOM platform. Two experienced investors were willing to join us and share their know how. 

In this news article we present you some selected ideas that were shared. If you are looking for full experience, you should watch recorded meeting on our website https://dex-ic.com/ce-connector_en.

You should also follow us on https://www.linkedin.com/company/dex-innovation-centre-dex-ic to be part of our future events.

Antonín Král – A brief overview of trends in AI

• Industry is shifting towards PyTorch and we can expect a even bigger increase once more open-source projects shift to it.
• We’ve seen a huge advancement in language processing/understanding in the last year (GPT-3, T5, BERT…)
• Moreover, federated learning has exploded in the last few years, e.g., predictive keyboards on Android. Which has risen a few privacy concerns.
• Deep Reinforcement Learning has shown a great number of possibilities, since advancements have been made and we can see a lot of possible applications, apart from playing games, possible uses include robotics and logistics coordination.
• One of the big market trends right now is AI application in Real-Time Analytics and IoT, which could help with smarter manufacturing and logistics (no more overstock)
• 5G could prove useful for IoT and other machine learning uses, just because of the huge amount of raw data, that will probably be necessary, if it expands to wider use.

Aleš Ducháč - Investing in AI start-ups

• AI start-ups are riskier than other investments
•  “Good” start-ups are not on the market for that long, they get invested in quickly. But watch out for FOMO!
• Since AI and machine learning is currently popular, the valuation can get high quickly, even with little to no traction.
• If a start-up claims to use AI, get them to explain to you if the AI is going to be useful, some of them might have integrated the AI just for being able to claim they have done so.
• If you think the start-up does not seem to have the needed knowledge for having a useful AI, consider getting someone to go through their code and assess what is there.
• Specific use cases you might want to check out for inspiration in what has worked/succeeded are: Ezra (MRI evaluation), Resistant AI (systems against manipulation and circumvention by advanced fraudsters), UiPath (robotic process automation of repetitive tasks)

The 2BeAngel meeting is brought to you by the CE-Connector initiative that brings together a group of private business angel investors with public fund support to co-invest in start-ups.