Passage 1 Artificial speech Because speech is the most convenient form of communication, in the future we want essentially natural
conversations with computers. The primary point of contact will be a simple device that will act as our window on the world. You will simply talk to it.
The device will be permanently connected to the internet and will beep relevant information up to you as it comes in. Just howquicklypeople willadapt to
a voice-based internet worldisuncertain. Many believe that, initially at least, we will need similar conventions for the voice to those we use at present on
screen: click, back, forward, and so on. But soon you will undoubtedly be able to interact by voice with all those IT-based services you currently connect
with over the Internet by means of a keyboard. This will help the Internet serve the entire population, not just techno-freaks. Changes
likethiswillencompass thewholeworld.Because English is the language of science, it will probably remain the language in which the technology is
mostadvanced, butmostspeech-recognitiontechniques aretransferable to other languages provided there is sufficient motivation to undertake the work. Withintenyearswewillhavecomputersthatrespondtogoal-directed conversations, but for a computer to have a conversation that takes into account
human social behaviour is probably 50 years off. We’ renot going to be chatting to the big screen in the living room just yet. There are those in the IT
community who believe that current techniques will eventually hit a brick wall. Personally, I believe that incremental developments in performance are
more likely. But it ’tsrue that by about 2040 or so, computer architectures will need to become highly parallel if performance is to keep increasing.
Perhaps that will inspire some radically new approaches to speech understanding that will supplant the methods we ’ re developing now. Small vendors
engage in simple spot-market transactions, with buyers and sellers dealing face to face to trade fairly standard products whose quality is easy to verify.
Lange multinational firms exchange more differentiated products, face greater difficulties in verifying quality, and must span greater separations in time
and space between one part of the transaction and the other. Most economies have both types of markets. But the first is relatively more common in
developing countries, the second in industrial economies. Developed markets- more global, more inclusive, and more integrated-offer more opportunity
and choice. Underdeveloped markets, more likely in poor countries, tend to be more local and segmented. So, compared with farmers in Canada, poor
farmers in Bangladesh have fewer opportunities and far fewer formal institutions (such as banks and formal courts ) to reduce their risks and increase
their opportunities. What limits market opportunities? The transaction costs stemming from inadequate information and incomplete definition and
enforcement of property rights. And barriers to entry for new participants. What increases them? Institutions that raise the returns from market
exchange, reduce risk, and increase efficiency. Yet not all institutions promote inclusive markets. Institutional designs that evolve through historical
circumstances or are directed by policy makers are not necessarily the best for all of society or for economic growth and poverty reduction. For
instance, state agricultural marketing boards, instead of helping farmers, have often resulted in lower incomes for them in Africa. And institutions that
once supported market transactions can outlive their usefulness, for example, privatization agencies and bank restructuring agencies. The challenge for
policymakers is to shape institutional development in ways that enhance economic development.

---