DailyPost 1052
ARTIFICIAL INTELLIGENCE RESEARCH BLUES!
Cost benefit analysis and return on investment have been both boon and curse for the IT industry depending on which angle you are looking from. One way you get the Q1 to Q4 cycle running profitably but on the other hand companies are missing out on the broader trends and later turn out to be miserable failures. The role of R&D, product development and its scalability become the crucial existential questions. The answers are few and far between. The losses incurred by the flagship of the AI world DeepMind to the tune of 470.2 million pounds in the year 2018 has brought the issues raised earlier to the mainstream public debate.
These losses are on the back of 302.2 million pounds in 2017 and a debt of 1.04 billion pounds due this year. Alphabet Inc.’s realisation of the importance of this company to the future of mankind is extremely encouraging. Dealing with real world challenges through R&D can only be the game changer. This streak is clearly evident in DeepMind. Alphabet’s spokesperson said in a statement, ”We will continue to invest in fundamental research and our world-class, interdisciplinary team, and look forward to the breakthroughs that lie ahead.”
The cost of research and the likelihood of low lying fruits may decide the future of AI research is a forewarning. DeepMind paid $483 million to approx 700 employees averaging $700,000. DeepMind’s StarCraft playing AI model consisting of 18 agents, had a training cost of $774,000. Other reinforcement learning projects will have similar costs. With this nature of cost of research, the academic world would most likely be left out in AI research and in the process open generic AI just might not happen, so necessary for mankind.
The fallout of the present scenario would be costly commercial AI products, not open to modifications, black boxes created by the commercial entities. ”The disastrous state of social media and addictive tech shows what happens when tech companies decide to give priority to their bottom line. At this crossroad, whatever is decided for the nature of AI research, will decide its course and also its impact.
RESEARCH CANNOT BE MADE TO FIT IN THE RETURN ON INVESTMENT FORMULA.
Sanjay Sahay